HighLoad++ 2023
2023-11-27..2023-11-28
-- online --
[HighLoad++ 2012, HighLoad++ 2013, HighLoad++ 2014, HighLoad++ 2015, HighLoad++ 2016, HighLoad++ 2017, HighLoad++ 2019, HighLoad++ 2022, HighLoad++ 2023, HighLoad++ 2024, HighLoad++ 2025]
| title | author | length |
|---|---|---|
| Yandex Internal Payment System: What's Under the Hood? | Anton Kuranda | 51m |
An overview of Yandex’s internal payments platform called FinOps, a separate system (not the old Yandex.Money) that handles payments across Yandex services… | ||
| The Evolution of Services and Development Tools | Sergey Berezhnoy | 38m |
A historical overview of how online services and developer tools have evolved in Russia from the early internet (1997) to the present, tracking growth from… | ||
| VKontakte Speech Recognition Engines | Ali Safiullin | 38m |
Describes VKontakte’s speech recognition engines, comparing online and offline systems by naming the neural network architectures used and explaining how they… | ||
| Failure Points in High-Load Systems: Backend | Konstantin Kozlovskiy | 39m |
Covers how backend systems fail under high load by categorizing common failure points (memory exhaustion, thread pool issues, etc.) and examining how service… | ||
| Dark Gods of Enterprise Architecture: Tales from the Warp's Depths | Andrey Zhukov | 40m |
Explains why different approaches to corporate architecture (design documents, various organizational patterns) still produce the same implementation problems… | ||
| Angie Web Server One Year Later: New Features and Plans | Valentin Bartenev | 46m |
A team of former nginx engineers founded a company in 2022 and began developing Angie, an open-source Russian web server. The talk reviews Angie one year… | ||
| Vulnerabilities in the Hyperledger Fabric Platform | Ilya Druzhinin | 41m |
Analyzes security issues in Hyperledger Fabric, detailing vulnerabilities in its consensus implementations (Raft, Kafka, SmartBFT) and platform architectural… | ||
| The Evolution of Go: How (Not) Our Reality Has Changed | Dmitriy Matrenichev | 31m |
Overview of recent and upcoming changes in the Go programming language over the past two years, focusing on concrete new features being added to the language… | ||
| No Time to Explain, Just Code! | Kirill Shvakov | 44m |
Talk describes why and how a developer created a new programming library when existing ORMs, JSON parsers and log parsers proved inadequate. It covers… | ||
| Alisa Six Years On | Pavel Kaplya | 44m |
A retrospective on the six-year evolution of Yandex Alice, a large-scale voice assistant serving tens of millions of monthly users. Covers challenges of… | ||
| How to Go Open Source and Not Lose Your Mind: YTsaurus Experience | Andrey Rivkin | 49m |
YTsaurus, Yandex’s big-data processing platform, is presented as a case study for open-sourcing an internal technology. The talk covers practical steps and… | ||
| Real-Time Face Recognition: Training Fast and Accurate Models | Vadim Selyutin | 33m |
Addresses building fast, accurate face-recognition models for real-time use on resource-limited devices by combining large deep-learning face models with… | ||
| Stop Writing Your Own Authentication! | Irina Blazhina | 44m |
Discusses why developers should stop building custom authentication and instead use existing solutions; covers authentication problems, risks of homegrown… | ||
| Total Employee Surveillance via Telegram | St. Zhukovskiy, Vasiliy Shalimov | 47m |
A presentation about an automated employee photo-reporting system for a large national beauty-salon chain (up to 20,000 staff). It explains using Telegram to… | ||
| Mattermost on Steroids: How We Launched and Evolve a Messenger | Evgeniy Kulgin, Yuriy Vlasov | 31m |
Describes challenges encountered while deploying and operating the Mattermost corporate messenger and how the team addressed them. Covers practical experience… | ||
| C++ and Security: Can We Do Better? | Sergey Talantov | 43m |
Examines claims from an NSA guide that C/C++ are ‘unsafe’ and should be replaced by languages like C#, Go, Java, Ruby, and Swift, and evaluates whether C++… | ||
| BARSiC — Asynchronous Replication and Consensus for 70 Databases | Grigoriy Buteyko | 43m |
Describes BARSiC, an asynchronous replication and consensus solution built for VKontakte’s infrastructure of 70 specialized database engines across ~800… | ||
| How and Why to Write Your Own Plugins for GoLand | Aleksandr Kirsanov | 50m |
Shows how to write custom plugins for GoLand and why to do it, based on VKontakte’s experience. Describes specific plugin development for the JetBrains GoLand… | ||
| What's Behind Generics in Go | Ilya Gorkun | 28m |
Explains Go generics: why Go’s generic design looks the way it does, how it differs from generics in other languages, and what implementation details like “gc… | ||
| Building an Anomaly Detector for High-Load from Python and Duct Tape | Evgeniy Rossinskiy | 45m |
Builds an anomaly detector for high-load services using Python and simple tools. Describes collecting many metrics (RPS, response time, open sockets, etc.)… | ||
| Communities Around a Technology: Why Being Free Is Not Enough | Kseniya Romanova | 36m |
How open-source projects succeed by building communities: the talk covers why being free is not enough and describes using Developer Relations (DevRel) tools… | ||
| YDB Topic Service: How We Improved Message Queue Performance | Aleksandr Zevaykin | 46m |
YDB Topic Service is a message-queue component within the YDB platform that provides durability, scalability, ordering, and delivery guarantees. The talk… | ||
| Too… Much… Async… What to Watch for When Working with a Feature | Viktor Burtsev | 40m |
Covers handling large numbers of asynchronous tasks in client-server systems when a new feature generates tens of background jobs. Discusses practical… | ||
| How to Effectively Rank All of Runet's Products | Aydar Gilazhev | 51m |
Describes Yandex's product search runtime and machine-learning components for ranking the entire Russian e‑commerce catalog. Covers the system and algorithms… | ||
| Did You Know What You Just Counted? Automated Verification of Numerical Computations | Igor Netay | 44m |
Describes problems with floating-point computations where every operation introduces rounding errors that accumulate and can invalidate results. Presents… | ||
| Interservice Authorization in Avito.PaaS | Anton Gubarev | 42m |
Describes implementing inter-service authorization in Avito.PaaS using Open Policy Agent (OPA) and Rego policies for over 2,000 services. Covers the… | ||
| A Brief History of NLP: From T9 to ChatGPT | Dani El Ayyass, Aleksandr Abramov | 49m |
A concise historical overview of NLP tracing developments from early mobile text input systems like T9 through to modern large language models such as ChatGPT… | ||
| The Recipe for Perfect Annotation in Computer Vision | Karina Kvanchiani, Aleksandr Kapitanov | 34m |
Describes creating and releasing three large open datasets for computer vision—HaGRID, EasyPortrait, and Slovo—covering different CV tasks. Explains the… | ||
| Code Generation and How to Use It Effectively | Aleksandr Sinichkin | 38m |
Explains practical use of code generation for high-load system development, focusing on why developers should not fear or avoid it and how to get the most… | ||
| Squeezing Maximum Performance Out of Go | Nikita Galushko | 40m |
Practical guidance for getting maximum performance out of Go programs, covering concrete techniques and pitfalls such as differences between various for-range… | ||
| China's Open Source Ecosystem: Past, Present and Future | Oleg Sirotyuk | 45m |
A historical and practical overview of China’s open-source ecosystem from 2000 to 2022, documenting how Chinese companies adopted and then became the world’s… | ||
| Transactional Replication in YTsaurus | Ruslan Savchenko | 39m |
Describes how transactional replication is implemented in YTsaurus, a distributed key-value store open-sourced by Yandex in 2023. Covers replication mechanisms… | ||
| The Evolution and Myths of CQRS | Andrey Tsvettsikh | 49m |
A developer with long experience in CQRS reviews its evolution and common misconceptions, comparing different interpretations of what CQRS means. The talk… | ||
| How We Built Ad Moderation from Scratch and Reached 1 Billion Verdicts per Day | Dm. Latyshev | 39m |
Describes building an ad moderation system at Yandex to process over 1 billion verdicts per day with automated checks on the order of single-digit seconds… | ||
| Data Sketches — How to Eat an Elephant Whole (Even If It's Infinite) | Sergey Zhemzhitskiy | 47m |
Introduces data sketching techniques for scalable data processing and analysis, focusing on algorithms like HyperLogLog, Bloom filters, and count-min sketches… | ||
| How We Manage Traffic for Thousands of Pods in a Multicluster Kubernetes Environment | Maksim Chudnovskiy | 38m |
Describes how to provide transparent pod-to-pod networking across many Kubernetes clusters so they behave like a single large “supercluster.” The solution uses… | ||
| How We Launched YandexGPT | Roman Gorb | 35m |
A talk describing how YandexGPT was taken from pretraining to product release, outlining the stages of model development and the practical problems… | ||
| How CV Helps Users Find Their Dream Product by Visual Example | Dmitriy Malakhov | 35m |
A presentation describes building a product that helps users find desired items by visual similarity using a question-driven interaction inspired by the… | ||
| Working with Arenas — Nearly Eliminating GC | Maksim Goroziy | 49m |
The talk explains using region-based memory management (arenas) in Go to reduce garbage collector overhead. It covers why arena allocation helps, practical… | ||
| How to Teach a Service to Report Errors So They're Clear to Users and Machines | A.Lyrchikov | 34m |
A talk about improving error reporting in a Go-based data storage system: it explains why generic “Something went wrong” banners were insufficient, examines Go… | ||
| Inside S3 | Pavel Levdik | 44m |
Description of Yandex’s S3-compatible storage that holds billions of files, focusing on the challenges of managing very large volumes of data and metadata… | ||
| Experience Evaluating Leading Chinese FOSS Solutions for Building IT Infrastructures | D.Varenov, V.Pavlov | 53m |
A review of advanced Chinese open-source operating system (OSS) solutions that are not included in mainstream Linux distributions, examining their technical… | ||
| Avito.AutoUpload: From 4M to 80M Active Listings | Evgeniy Tolmachev | 37m |
Description of Avito’s Auto-upload system that automates clients’ management of classified ads; the system is implemented as a set of microservices and is one… | ||
| Anyone Can Generate Load | Kseniya Biryukova | 48m |
Presentation argues that many companies skip load testing because it requires specialists and specific tools, and describes practical approaches and tools to… | ||
| Firewall in the Cloud: Ways to Integrate into Network Architectures | Nikita Razhev | 52m |
Describes techniques for deploying firewalls in cloud network architectures, comparing different insertion methods and routing setups. Explains an evolution… | ||
| Prompt Engineering: The Path to Effective Work with ChatGPT | Ilya Kozyrev | 01h 56m |
Covers how large language models work and the central role of prompts when interacting with ChatGPT, focusing on practical prompt-engineering techniques… | ||
| VK Media Recommendations: How We Build the Personalized "For You" Feed | S.Malkevych | 41m |
Describes how VKontakte built a new personalized “For You” media recommendations feed, focusing on machine learning, analytics, and backend engineering. Covers… | ||
| How to Survive a One-Year Sprint and Migrate to ClickHouse | Denis Trayzhon | 51m |
A presentation about migrating a product or startup from existing analytics storage to ClickHouse, describing practical challenges encountered during a… | ||
| How We Saved Cloud Budget by Using Autoscaling | V.Vasilev, V.Magsumov | 42m |
A case study of a recommendation platform built with state-of-the-art ML algorithms that consumed very large compute (example: ~930 CPUs and 4.7 TB RAM per… | ||
| Deep dive into the Go scheduler, or Why would I steal goroutines? | Andrey Shchipilo | 49m |
A technical deep-dive into Go’s scheduler explaining why and how goroutine stealing (work stealing) works inside the Go runtime. Covers Go runtime internals… | ||
| FerretDB — mongoDB on the outside, PostgreSQL on the inside | Maksim Kubarskiy | 43m |
FerretDB is an open-source Go proxy that accepts MongoDB wire protocol requests and translates them into SQL for PostgreSQL (using jsonb), providing… | ||
| Myths and realities of multi-master architecture in PostgreSQL | Mikhail Zhilin, Pavel Konotopov | 53m |
Examines multi-master architecture in PostgreSQL by addressing common myths (that it is slow, hard to integrate with applications, and difficult to maintain)… | ||
| (Un)successful experiment in growing an Open Source culture | Oleg Balbekov | 46m |
An exploration of building a business around open source: describes practical benefits (hiring, content for articles and talks) and recounts an experiment in… | ||
| Scaling Yandex search throughput by 4× | Sergey Skorodumov | 47m |
Describes how Yandex increased search throughput fourfold by adapting to hardware loss and supply constraints; they analyzed bottlenecks and changed system… | ||
| Under the hood of fast traffic splitting for A/B testing | Dmitriy Volkov | 39m |
Explains the internal design of a high-performance traffic splitting system for A/B testing, focusing on how to route users to experiments with low latency and… | ||
| Becoming your own vendor: Deploying EVPN in VK Cloud | Aleksandr Popov | 48m |
VK Cloud implemented EVPN to provide VLAN-agnostic Layer 2 connectivity between its cloud and customers’ on-premises data centers. The talk explains the… | ||
| Architecture of Yandex Market’s infinite personal feed | Vladimir Bondarevskiy | 43m |
Describes the architecture behind Yandex.Market’s infinite personalized feed launched on the app home page. Covers the performance and scalability challenges… | ||
| Implementing OLAP: how we built columnar storage in YDB | Sofya Novozhilova | 52m |
YDB is an OLTP platform and the talk explains how columnar storage was implemented on top of it to support OLAP workloads. It describes the technologies used… | ||
| Arc — Yandex’s internal VCS for a monorepo | Stepan Polokhin | 46m |
Yandex built an internal version-control system called Arc to manage its huge monorepo. The talk explains problems with standard VCS tools at that scale… | ||
| Peculiarities of developing an Open Source app for real-time IP camera streaming | Aleksey Khit | 43m |
A developer describes the challenges of building an open-source real-time IP camera streaming application called go2rtc using the Go programming language. The… | ||
| Sharding Postgres without doing it yourself | Denis Volkov, Kirill Reshke | 42m |
A production-ready, open-source stateless Postgres query router for horizontal scaling via sharding is presented. It implements the Postgres protocol, is… | ||
| A talk on the Digital Ruble | Vyacheslav Adamov | 50m |
Overview of the Digital Ruble: why central banks worry about private digital currencies and specific risks for the national currency and payment system demand… | ||
| How Java is created | Oleg Chirukhin | 51m |
A talk examining how large open-source Java projects are organized so they avoid chaos and produce finished, working products. It discusses project practices… | ||
| Reliable analytics: how we collect stats in VK mini-apps | Oleg Mifle | 39m |
Describes how VK Mini Apps collect reliable analytics across web, iOS, and Android clients. Covers the challenges of gathering accurate usage and performance… | ||
| From CRM to DataLake with K8s and microservices | Andrey Vilmov | 38m |
Describes migrating a growing CRM into a Data Lake architecture using Kubernetes and microservices to handle integrations with multiple internal and external… | ||
| How we used event sourcing to organize data delivery and schema updates | Nikita Ryanov | 52m |
Describes using event sourcing to deliver structured data into a storage while allowing table schemas to evolve. Explains challenges of schema change… | ||
| How our systems gain resilience, and why failures are useful | Sergey Reusin | 49m |
The talk explains how modern high-load systems are made resilient by embracing and managing failures rather than avoiding them. It discusses conflicts that… | ||
| How Nomad can shoot you in the foot, and what to do to dodge it | Vladimir Radonets | 36m |
Compares running workloads on Kubernetes versus HashiCorp Nomad, explaining risks and mistakes teams make when switching to Nomad. Covers operational problems… | ||
| Integrating GigaChat LLM into a virtual assistant: technical implementation | D. Antyukhov, P. Gladkikh | 41m |
Describes integrating the GigaChat large language model into Sber’s virtual assistant to enhance its capabilities; covers why an LLM is needed in the assistant… | ||
| Scalable big data processing pipeline with Cadence | Aleksandr Vanyushkin | 32m |
A scalable big-data file-processing pipeline built using Cadence (Uber’s workflow engine) to handle large files, failures and peak loads. The talk describes… | ||
| database/sql: the bad, the good and the ugly — developing a driver for YDB | Aleksey Myasnikov | 45m |
A talk about implementing a Go database driver for YDB using the standard library package database/sql. It covers historical evolution of database/sql, design… | ||
| Mathematical highload: big, very big, and unimaginably large numbers | Aleksandr Kirsanov | 42m |
A mathematical perspective on what “big” really means: compares common IT scales (billions of users, terabytes) to much larger mathematical magnitudes and… | ||
| Open Source AppSec Review: how to perform acceptance, deployment and hardening of a solution | L.Khakimov, A.Fedulaev | 42m |
A practical guide to evaluating, adopting, and hardening open-source application security solutions, focusing on processes for acceptance (review), deployment… | ||
| Distributed tracing with Jaeger and ClickHouse | Filipp Bocharov | 38m |
Describes building an internal distributed tracing service for a large MTS product ecosystem using Jaeger for span collection and ClickHouse for storage and… | ||
| LLMops: what's beyond ChatGPT and how you can deploy it | Irina Nikolaeva | 49m |
Overview of LLMops beyond ChatGPT: explains basic ML terminology, reviews classic MLOps principles, and examines why large language models are powerful… | ||
| Evolution of a data platform for timely communications with 100+ million customers | Ilya Nechaev | 39m |
Describes building a data platform to support timely communications with over 100 million customers, explaining the business goal that requires data and the… | ||
| This is all that will remain after me: challenges of code inheritance and transferring ownership | I. Shakhtarina | 50m |
Talk discusses legal and practical issues around code inheritance and transfer of rights, explaining how intellectual property and ownership of software are… | ||
| How the Spectro robot scanning system works | Valeriy Ilin | 46m |
Description of the Spectro robot’s scanning system focusing on computer-vision components, the performance challenges encountered during development, and how… | ||
| Scripts in applications: how and why to let users write code? | Andrey Privalov | 42m |
Covers enabling users to write and run scripts inside applications, with examples like nginx plugins in Lua and Traefik plugins in Go. Explains why embedded… | ||
| Boilerplate as a tool for standardizing Go projects | Aleksandr Kaloshin, Konstantin Pastor Gerte | 44m |
Describes using a Go boilerplate to standardize microservice projects by providing reusable, common packages and reducing repetitive code. Covers tooling and… | ||
| Sharding: from scratch to Yandex Disk | Andrey Kolnoochenko | 49m |
A detailed look at how Yandex.Disk’s database sharding evolved: the talk covers the architecture changes, the operational and design problems encountered, and… | ||
| One pipeline to rule them all | Evgeniy Kharchenko | 49m |
Demonstrates using a single CI/CD pipeline to share expertise across teams, speed up team delivery, and introduce DevOps practices. Describes concrete… | ||
| Replacing four monitoring systems in four years and surviving | Aleksandr Kachmashev | 42m |
Describes replacing four different monitoring systems over four years for a high-load production environment, naming the cause-and-effect failures (five key… | ||
| Zero-cost I/O and fault tolerance in distributed deep learning | Aleksey Morozov | 43m |
Describes how Yandex built a zero-cost infrastructure for distributed deep learning on top of a distributed transactional file system, enabling distributed… | ||
| Extending GDB debugger capabilities with Python | Mikhail Elkhimov | 50m |
Shows how to extend the GDB debugger using Python to improve debugging of production issues such as crashes and core dumps. Covers using GDB Python extensions… | ||
| How to build a container without arming a hacker | Anton Zhabolenko, Aleksey Fedulaev | 48m |
Examines container security risks from Living off the Land (LotL) attacks, where attackers reuse legitimate utilities inside containers. Covers which common… | ||
| Generative language models | Aleksandr Proshunin | 50m |
Overview of the evolution of natural language processing from early probabilistic models to modern large language models, discussing notable architectures and… | ||
| How to test Go code with a database? | Vasiliy Vasilyuk | 39m |
Shows how to test Go code that interacts with a database by reducing reliance on fragile mocks and making tests easier to maintain as the codebase evolves… | ||
| Protobuf and buf: glamour, misery, and import substitution | Edgar Sipki | 37m |
A talk compares Protocol Buffers (protobuf) and the buf tool, examining their strengths and weaknesses for efficient, compatible data serialization in… | ||
| Kubernetes without internet | Maksim Nabokikh | 41m |
Practical guide to running Kubernetes in completely air-gapped environments (banks and critical information infrastructure) where no Internet access is… | ||
| Data bus features for very large installations: the YDB Topics example | Aleksey Dmitriev | 40m |
YDB Topics is presented as a data bus solution for very large installations (thousands of servers), focusing on the specific requirements and behavior such… | ||
| Modern methods for building a monitoring platform | Ilya Petropavlovskiy | 29m |
Overview of modern approaches to building monitoring platforms, comparing classic waterfall, iterative/Agile and a centralized (“state”) approach, with… | ||
| Why build resource-hungry software: principles of the reconciliation loop (hello, K8s!) | Georgiy Melikov | 46m |
Explains why building resource-hungry (“greedy”) software can be beneficial and describes the reconciliation loop pattern used by systems like Kubernetes… | ||
| Fast data processing in a data lake with SQL | Vladimir Ozerov | 51m |
Describes using distributed SQL engines—Trino, Presto and Dremio—to run SQL queries directly over files in a data lake, reducing ETL and enabling more… | ||
| Exception Handling: through multiverses of interoperability | Maksim Kokryashkin | 39m |
Covers exception handling across the Lua and C/C++ interoperability in the Tarantool in-memory platform. Explains how errors propagate between Lua and C/C++… | ||
| BrainSploit: social engineering exploits | Anton Bochkarev | 48m |
BrainSploit examines social-engineering exploits used in socio-technical penetration testing, comparing modern project scenarios to classic cases like Kevin… | ||
| What fighting spammers in Beeline Anti-Fraud looks like through a Data Scientist's eyes | Nataliya Bagrova | 48m |
Beeline’s Antispam team (part of Anti-Fraud) describes building a service to protect subscribers from unwanted and advertising spam calls. The talk covers… | ||
| High-load Open Source VDI for the CyberCamp cyber range platform | Semen Baryshnikov | 31m |
An open-source VDI (virtual desktop infrastructure) solution built to run the CyberCamp cyber-exercise/CTF platform. It covers creating isolated team… | ||
| From pytest to Go: a fixture-based test environment | Timofey Kulin | 41m |
Describes porting the pytest fixture approach into Go to make tests concise by handling environment setup via fixtures. Covers how fixtures are implemented in… | ||
| Magnit Tech: Inventory and Pricing Services in Go — Handling Large Data Streams | D.Zhavoronkov | 37m |
Describes building Magnit Tech’s inventory and pricing system in Go to handle very large data volumes (about 30k stores, 10–20k SKUs, ~500 million rows) and… | ||
| Building Resilient Application Deployments on Kubernetes | Oleg Voznesenskiy | 44m |
Describes building fault-tolerant deployment pipelines for applications running on Kubernetes, focusing on reducing failures caused by change rollout (citing… | ||
| Compute/Storage Separation in Greenplum | Andrey Borodin | 41m |
Yezzey is an open-source GreenplumDB extension that moves a table’s storage to S3 while preserving the database’s native on-disk data format. The talk shows… | ||
| How to Build Binary-Compatible APIs in Compiled Languages? | Aleksandr Turenko | 40m |
Describes how to design binary-compatible APIs for compiled languages when splitting a product into independently evolving components. Covers problems like ABI… | ||
| zkSNARKs — Cryptographic Proofs for Scaling and Security Challenges | Sergey Prilutskiy | 48m |
Explains zkSNARKs as a cryptographic tool for scalability and security, describing how succinct zero-knowledge proofs can be used in scaling services and in… | ||
| Hadoop in the Cloud: The Story of Migrating Hundreds of Petabytes | Mikhail Maryufich | 45m |
Migration of OK’s massive Hadoop platform (over 200 PB in HDFS, ~70k vcores and ~200 TB RAM) into the cloud is described, covering both analytics and… | ||
| Managing User Expectations of Your Platform | Viktor Popov | 48m |
Examines why users become dissatisfied when a platform functions correctly but does not match their expectations, identifying gaps between user mental models… | ||
| New Opportunities in HR Tech: Solving Generative Tasks with Transformer+LoRA+RLHF | Mark Panenko | 45m |
Describes using Transformer-based large language models for HR tech generative tasks by fine-tuning with LoRA adapters and incorporating human feedback via… | ||
| Migrating a Data Mart from Teradata to Greenplum | Stanislav Sviridenko | 36m |
Describes migrating a data warehouse from Teradata to Greenplum, covering challenges beyond code conversion and data transfer such as subtle behavioral… | ||
| 200 Integrations with 5 Developers | Denis Bozhok | 43m |
Describes building a single unified interface to manage hundreds of external integrations (payments, fuel stations, hotel sales) using a small engineering team… | ||
| Fuzzing Testing: Practical Applications | Nikita Dogaev | 30m |
Fuzzing testing: practical techniques for generating and using random input data to find bugs, contrasted with ordinary unit tests that use a simple… | ||
| From CPU to GPU: When Cores Decide | Vitaliy Shutov | 50m |
The talk explains when and why to move workloads from CPUs to GPUs, comparing CPU and GPU architectures and performance trade-offs. It covers GPU use cases in… | ||
| How to Invest in Open Source Without Burning Out? | Anton Vinogradov | 48m |
Presentation explains how a company builds commercial services around an open-source distributed database (Apache Ignite) and how to invest resources in open… | ||
| Hazelcast That Could: Getting Lightning-Fast Responses at Hundreds of RPS | Prokhor Krylov | 38m |
Describes using Hazelcast in a distributed, service-oriented system to serve hundreds of user requests per second with near-instant responses by avoiding… | ||
| Use the Force, Luke: Single Pane of Glass in the World of SRE | Kirill Borisov | 48m |
Describes the “Wall of Dashboard” anti-pattern in observability for SRE: too many dashboards causing information overload and loss of focus. Explains problems… | ||
| Training a Bank Support Bot: Quality Data and Thousands of Intents | Irina Stepanyuk | 32m |
Tinkoff’s support bot Oleg is trained with NLP models to handle over 1,000 distinct intents across products and customer-service topics. The talk covers… | ||
| Internal Developer Platform and Developers: What Does the Business Pay For? | Aleksandr Chubinskiy | 54m |
Discussion of building and operating an internal developer platform for a large, fast‑growing company, focusing on which platform features and services the… | ||
| Current Security Threats in Large Language Model Applications | Artem Bachevskiy | 47m |
Overview of current security threats affecting applications that use large language models, covering risks in LLMs themselves and in systems built on them… | ||
| Petabytes in YDB over HDD in Yandex.Metrica Processing | Anton Barabanov | 42m |
A case study of building a petabyte-scale YDB (Yandex Database) storage layer on HDDs for Yandex.Metrica’s mobile analytics system, describing architectural… | ||
| Domain-Driven Design in Go — It Doesn't Hurt (Much) | Ilya Sergunin | 47m |
Shows how to apply Domain-Driven Design in Go by breaking DDD into practical elements and patterns: ubiquitous language, layered architecture, self-validating… | ||
| How to Become a Senior Engineer | Vasiliy Romanov | 50m |
Examines what Go (Golang) knowledge is required to be a senior engineer: compares senior vs other grades, lists senior anti-patterns, and shows that seniors… | ||
| 5 New Ways to Use Data in Your Kafka | Andrey Serebryanskiy | 44m |
Practical techniques for using Kafka data in five new ways: building per-second updating dashboards, real-time fraud detection, and fast reliable rendering of… | ||
| How Free Projects Are Developed on the ALT Team? | Evgeniy Sinelnikov | 43m |
Overview of how the ALT Linux Team organizes development of the Sisyphus free-software repository: the talk describes the team’s open development workflow… | ||
| YDB Performance Optimizations for ARM | Maksim Kita | 38m |
YDB performance optimizations for ARM are examined, describing the specific problems encountered when porting and tuning YDB for ARM architecture and the… | ||
| Evolution of the Transcoder Architecture | Dmitriy Lukshto | 37m |
Describes a video transcoding system that converts user videos into popular resolutions, computes video fingerprints, and generates subtitles. Explains how the… | ||
| NeuroHD: Hunting Jackals, or Deploying and Accelerating Networks in VK Infrastructure | Arseniy Nerinovskiy | 35m |
NeuroHD is VK’s video-enhancement technology that increases resolution, raises frame rate, and reduces noise. It runs using two neural networks deployed and… | ||
| How to Reach Consensus Using Paper, Pencil and the Raft Algorithm | Yaroslav Dynnikov | 01h 27m |
Explains the Raft consensus algorithm and how to teach and understand it using only paper, pencil, and manual simulation. Covers Raft’s use in distributed… | ||
| How to Make a Platform Delightful, or Avito PaaS After Five Years | Aleksandr Lukyanchenko | 47m |
Avito describes the evolution of its internal PaaS over five years, focusing on making the developer platform more convenient. The talk covers problems caused… | ||
| From Plank to Computer: The Path from Weaving to the Electronic Computer | Irina Shakhtarina, Pavel Arkhipov | 33m |
The talk traces technological continuity from hand weaving to early computers, explaining how innovations in loom design and weaving techniques—especially… | ||
| Pattern Search on String Sequences in a Database | Evgeniy Zverev | 45m |
Describes pattern search over sequences of strings stored in a database, with examples such as finding price-movement patterns for financial analytics and… | ||
| Go in Domain Driven Design | Dmitriy Anisov | 44m |
Covers applying Domain-Driven Design (DDD) to backend services written in Go, explaining why to use DDD, its advantages and drawbacks, and specific challenges… | ||
| Interviewing Senior Developers: Testing Soft Skills with Hard-Skills Questions | Ivan Matveev | 48m |
A talk about interviewing senior Go developers that argues soft skills can be assessed through technical (hard-skill) questions. It discusses typical Go… | ||
| Scaling OpenAPI in an API-First Company | Maksim Lapshin | 48m |
A Flussonic team describes adopting an API-first approach and scaling OpenAPI 3.1 across a polyglot backend (Erlang, Go, Python, Rust, Rails, C, Node.js) to… | ||
| Open Source for Companies and Developers: Boost, C++, and userver as Examples | Anton Polukhin | 47m |
Overview of open-source usage in companies and by developers using Boost, C++, and the userver framework; discusses the diverse scale and benefits of… | ||
| You Can't Build an NGFW on Linus Torvalds' Creativity | Denis Korablev | 40m |
The talk examines challenges of building a next-generation firewall (NGFW) for enterprise use, arguing that ad-hoc, hobbyist-style development (à la Linus… | ||
| The Thorny Road of a Digital Design Tool | Viktor Vyskrebentsev | 48m |
A practitioner describes building a digital design tool for architecture and integration in a large organization, explaining why C4 models and simple… | ||
| Clean Labels for ML | Anastasiya Bezzubtseva | 42m |
Examines how the quality of ML models depends on the labels used for training and practical methods to improve labels collected from crowdsourcing platforms… | ||
| High Throughput and/or Low Latency | Maksim Yunusov | 49m |
A case study of reworking a real production system to multiply throughput on existing hardware while also reducing response latency. It describes diagnosing… | ||
| The Index Structure in Mail.ru Mail | Rustem Gafarov | 46m |
Explanation of Mail.ru’s email indexing architecture, describing how mail messages are indexed for search and retrieval. Covers specific technical problems… | ||
| An Algorithm for Incremental Backups in Apache Ignite | Maksim Timonin | 47m |
Describes an algorithm for creating incremental backups (snapshots) in the distributed in-memory database Apache Ignite, addressing the additional challenges… | ||
| Trees in Interviews: Preparing for an Algorithmic Interview (for One Big Company) | Maksim Fatin | 01h 58m |
Covers common binary-tree problems asked in technical interviews at major Big Tech companies in Russia, focusing on preparation strategies and typical… | ||
| How We Developed the Core of the National Domain Zone Registry | Aleksey Kostyuchenko | 36m |
A presentation about developing a new registry core for Belarus’s national top-level domain (BY/БЕЛ). It covers the registry’s history, its core operational… | ||
| A Translator for a Language You Cannot Speak or Write | A. Kapitanov, A. Nagaev | 48m |
A presentation explains building a translator for a language that cannot be spoken or written, using Russian Sign Language (RSL) as the example and arguing… | ||
| Quality and Governance in a Large Open Source Project | Aleksandr Makarov | 01h 14m |
A large open-source project matured and became widely used, and the talk examines maintaining and evolving it after rapid growth. It discusses quality control… | ||
| MaaS — Monitoring as a Service | Valentin Lebedev | 42m |
Describes Monitoring as a Service (MaaS) for high‑load systems: using monitoring to detect service problems before customers report them and to derive… | ||
| How to Practically Optimize Cloud Spend with FinOps | Vyacheslav Bessonov | 45m |
Overview of practical FinOps for cloud cost optimization in Russia: describes available Russian-market FinOps tools, the implementation process for these… | ||
| How We Built a Web Framework from Scratch in 1C and Its Cloud Control Panel | Basel Darvish | 38m |
A Russian team built a web framework from scratch on top of 1C and used it to implement a cloud control panel. The talk explains the motivation—standardizing… | ||
| Eventual Consistency in a Stateful Service | Dmitriy Isaev | 52m |
A presentation about designing and operating a stateful distributed storage system of over 80 TB that faces scaling challenges and cannot provide strong… | ||
| The Story of an Incident: How to Paralyze Work for 20k Employees | Eduard Aleksandrov | 47m |
Postmortem of an incident in an internal Yandex Tracker deployment: how their incident-management pipeline detected and responded to a failure that paralyzed… | ||
| When You Need to Build Your Own Database | Aleksandr Biryukov | 41m |
Examines when it becomes necessary for a company to implement its own database rather than use existing solutions. Discusses the practical problems and… | ||
| Partial Object Modification in Yandex Object Storage: How We Improve Filesystem Behavior | Aleksandr Snopov | 39m |
Yandex Object Storage supports partial object modification to improve filesystem-like operations. The talk explains the problem of updating portions of stored… | ||
| Data Streams, Graphs, State Machines — Building Business Logic in Go Microservices | Apres Antonyan | 37m |
Describes building business logic inside Go microservices for a marketplace, focusing on data streams, graphs, and state machines as the core technologies… | ||
| Catch in 60 Seconds: Rapid Incident Response Using Behavioral Transformers | A. Goy | 44m |
Describes using behavioral transformers to detect and respond to account takeover and mailbox compromise quickly; explains applying transformer-based models on… | ||
| To Find Everyone and Bring Them Together: Event Tracing System in VK Calls | Aleksey Shpagin | 41m |
VK Calls implemented an event tracing system to diagnose call issues like failed connections, dropouts, and missing screen sharing. The system collects and… | ||
| Inspired by the Sixth Labour of Hercules | Dmitriy Kyrkhlarov | 40m |
A case-study describing how a mid-sized IT company eliminated 20 years of technical debt to build a disaster-tolerant production system. It details practical… | ||
| Operating cilium in VK Clusters | Mikhail Petrov | 42m |
Describes why VK adopted the Cilium CNI, compares it to prior CNIs used at VK, and explains how Cilium is operated in their Kubernetes clusters. Covers… | ||
| Building an OMS with Temporal: From Zero to Tens of Thousands of Orders per Day | Evgeniy Konechnyy | 47m |
Describes building an Order Management System (OMS) using Temporal to handle order processing in a microservices-based e‑commerce environment. Explains why… | ||
| SRE Practices as Managing an Apartment Building | Gleb Goncharov | 46m |
Compares IT infrastructure and SRE practices to managing an apartment building, framing both as processes aiming to keep systems running and satisfy end users… | ||
| Welcome to the Real World, Robot! | Aleksandr Chistyakov | 43m |
Covers methods for validating autonomous cars before city deployment, focusing on safety and efficiency testing. Describes improving test coverage of… | ||
| Does ChatGPT Upset Daniil Podolsky | Varvara Podolskaya, Daniil Podolskiy | 40m |
Examines whether ChatGPT can help someone pass a senior-level developer interview and how to use ChatGPT effectively for answering questions and generating Go… | ||
| ETL on Kafka + Confluent: Problems and Solutions with Go | Nikita Stepanchenko, Yura Sargsyan | 34m |
Describes building an ETL pipeline for product data using Apache Kafka, Confluent platform and kSQL to handle rapidly changing product information at peaks… | ||
| TechTalk "Just Like Everyone Else: Introducing Modern Engineering Practices into Legacy Systems" | Nikita Priymak | 13m |
An engineer describes a legacy loan-servicing system at Gazprombank and uses it as a concrete example to define what a legacy system is. The talk covers… | ||
| TechTalk "YandexGPT: How We Built It and What’s Next" | Petr Ermakov | 14m |
Yandex developed its own large language model called YandexGPT, and this talk explains why the company decided to build an in-house LLM and how the development… | ||
| TechTalk "Kubernetes in the Plat.Form World" | Aleksey Nazarov | 11m |
Organization, configuration, and maintenance of a Kubernetes platform used at Plat.Form, including how project teams are integrated into that Kubernetes… | ||
| TechTalk "Anti-Fraud Platform — Verification Technologies" | Maksim Petrov | 10m |
Overview of Beeline’s anti-fraud platform: business and CTO priorities, the platform’s technical foundation, and the tools used. Covers scaling and… | ||
| TechTalk "Not by Time Spent Alone: How We Build a Ranking System" | Anastasiya Pavlovskaya | 11m |
A talk describing how a recommendation and ranking system is built: it covers the system architecture, the need for post-ranking (re-ranking) even when the… | ||
| TechTalk "Open Source as a Contribution to Technological Evolution" | Sergey Berezhnoy | 11m |
Explains Yandex’s view of open source, why Yandex releases its internal technologies publicly, and how to organize open-source work inside a large corporation… | ||
| TechTalk "The Compromise Point Between Business and System Performance" | Anton Abramov | 10m |
Presentation examines balancing increasing business load with system performance and resilience using Gazprombank’s retail CRM suite as a case study. It… | ||
| TechTalk "How to Move Gigabytes of Data Out of an Isolated Network" | Natalya Azisova | 11m |
Shows methods for moving gigabytes of data out of an isolated network segment by designing integrations between network zones with different access levels… | ||
| TechTalk "Event-Driven Microservices and the Challenges of Migration" | Alena Melnik | 11m |
Covers event-driven microservice architecture: why teams adopt event-based designs, the risks of distributed systems, and specific load-related challenges when… | ||
| HighLoad++ 2023 Award for Contribution to the Development of the Internet Development Ecosystem in Russia | Aleksey Borovets | 01h 28m |
Announcement of the HighLoad++ 2023 award recognizing contributions to Russia’s web development ecosystem, tied to the HighLoad++ professional conference for… | ||
| HighLoad++ 2023 Opening | Oleg Nin | 22m |
Announcement for HighLoad++ 2023, a major professional conference for developers of high-load systems and Go developers, with events in Moscow (November 27–28… | ||
| Conference Closing | Aleksey Rovets, Oleg Bunin | 19m |
A closing-session announcement for HighLoad++ 2023 and related Go developer conferences, listing dates and venues (Saint Petersburg DESIGN DISTRICT DAA in June… | ||
| HighLoad++ 2023 Report | 03m | |
A report about the HighLoad++ 2023 conference, a professional event for developers of high-load systems and Go developers held in November 2023 in Moscow. It… | ||
| How We Do Tracing at the Scale of Thousands of Services and Millions of Spans per Second | Igor Balyuk | 51m |
A presentation about how Avito implements distributed tracing to handle millions of spans per second from several thousand services. It describes the problem… | ||
| nothing matches | ||