PyData Austin 2019
2019-12-07..2019-12-08
Austin, United States
| title | author | length |
|---|---|---|
| Extending Python Into the Future | Travis E. Oliphant | 15m |
A proposal to build ePython, a typed-subset extension language that compiles one source to multiple runtimes (CPython PyPy RustPython MicroPython Jython… | ||
| What the...? Data Science Questions Asked and Answered | Katrina Riehl | 24m |
A run-through of concrete claims and tradeoffs: a security company running services for 200 million Internet properties and using a lava lamp RNG to seed… | ||
| Becoming a Data Head, a reflection on the trade offs and decision | Andy Terrel | 48m |
Claims concrete tradeoffs: builds mission-driven hiring to recruit people, runs a platform data team not point-to-point, and accepts that ~50% project failure… | ||
| Q&A with the CEO of Anaconda | Peter Wang | 50m |
Origins story of a 2012 workshop that led to the creation of the Anaconda distribution and the package manager conda, a concrete build that largely works and… | ||
| Data Science How do you Even? | Julie Hollek | 33m |
Claims concrete wins: an implemented quality filter yields a 40% reduction in blocks following @mentions from accounts users don’t follow, a timeout… | ||
| Customizing JupyterLab using extensions | Alex Bozarth, Luciano Resende | 35m |
Shows how to install an existing extension (table of contents) via the extension manager, explains front-end vs server extensions, then generates a real… | ||
| Panel: Dashboards for PyData | James A. Bednar | 01h 26m |
Demonstrates building interactive Python dashboards with Panel (and mentions Voila), showing a one‑liner to get an app, how interact auto‑creates widgets and… | ||
| uarray: Separating interface from implementation | Travis E. Oliphant | 36m |
Claims that NumPy must separate interface from implementation and that __array_ufunc__ and __array_function__ are insufficient, and presents a working… | ||
| 3 Case Studies in Applying Advanced Technology... | Kevin Horecka | 46m |
Three concrete case studies describe what was built and what it showed: a health‑economics analysis that finds parvo treatment costs about $56 average parvo… | ||
| Q&A with Keynote Katrina Riehl | Katrina Riehl | 37m |
Recommends concrete steps for healthier projects: require affiliation with NumFOCUS or at least a code of conduct, add a first-contribution guide and follow-up… | ||
| Multilingual embeddings to scale NLP models to multiple... | Deeksha Yennam | 22m |
Shows that a prebuilt multilingual universal sentence encoder and Facebook MUSE actually align vectors across languages (MUSE for ~30 languages, Google encoder… | ||
| How Neuroscience Makes AI Intelligent | Dalton A. R. Sakthivadivel | 32m |
Argues concrete parallels: learning uses synaptic plasticity plus error minimization, with brains holding 80 to 100 billion neurons whose weights change via… | ||
| Addressing the Scarcity of Data Scientist | Joe Gartner | 43m |
Claims that immersive programs close the hiring gap by producing day‑one practitioners who use Python AWS scikit-learn Spark Postgres and SQL to build… | ||
| Frictionless Data Pipelines | Lilly Winfree | 38m |
Claims a working integration: Frictionless Data Python libraries get wired into oceanography ingest pipelines to clean and standardise real messy submissions… | ||
| Objectionable Content | James Powell | 01h 27m |
Clear claim: use the Python object model to build production systems (not ML), with concrete patterns and tradeoffs. Demos include a resistor-network builder… | ||
| ADTK: An open-source Python toolkit for anomaly detection in... | Tailai Wen | 44m |
ADTK claims to be an open‑source Python toolkit that supplies ready‑to‑use building blocks — detectors, transformers, aggregators — with a unified API so… | ||
| Speeding up Machine Learning tasks using GPUs in Python | Saloni Jain | 34m |
RAPIDS cuML implements GPU versions of Decision Trees and Random Forest with claimed 10x–50x speedup and an end-to-end demo on a 200 GB mortgage CSV running on… | ||
| Machine Learning Crash Course | Samuel Taylor | 42m |
Begins with concrete claims and concrete tools: builds classifiers and regressors using Python and scikit-learn with train_test_split, turns job titles into… | ||
| Q&A with Keynote Travis Oliphant | Travis E. Oliphant | 39m |
A candid account of how open source converts to cash claims 3,000 copies sold (~$90,000) with ~$60,000 funding a PhD student, argues that companies monetize… | ||
| Hyperrest: A new Apache Arrow API For High Performance Data... | Sudheesh Katkam | 37m |
Hyperrest is an Arrow-based API implemented in Dremio that replaces row‑centric ODBC transfer with Arrow Flight style columnar transfer so Pandas avoids… | ||
| Recommendations at Rappi: from MVP to Data Product | Ariel Wolfmann | 38m |
Describes a working pipeline that moves from simple baselines to a production recommender: start with random and top-sold lists, establish a baseline, then add… | ||
| Lightning Talks | 43m | |
A grab bag of concrete outcomes and tools not promises: you can start a DIY self‑driving pipeline in an hour by recording phone video and training on labeled… | ||
| Introducting Dask-Gateway: Dask clusters as a service | Jim Crist | 41m |
Dask-Gateway provides a secure, multi‑tenant server that lets users launch Dask clusters via a REST API and central admin policies, supporting backends like… | ||
| An Introduction to Sentiment Analysis of... | Fatma Tarlaci, Dhavide Aruliah | 01h 31m |
Shows a step‑by‑step claim that one can build an SA model from scratch with open‑source Python, specifically NLTK and spaCy alongside scikit‑learn; covers… | ||
| Values-based OSS Open Discussion | Peter Wang | 45m |
Argues that the real issue is concrete: cloud vendors relicense or slot their APIs under familiar names (example: TensorFlow reworking NumPy APIs, Amazon/Redis… | ||
| Evaluation of Cloud Hosting Frameworks for Machine... | Pushkar Kumar Jain | 41m |
Builds regression‑based anomaly detectors for alternator coil temperatures using pandas, scikit-learn, Keras and ADTK, serializes them as pickle files and… | ||
| Develop Data Science for Business Value: a Phoenix Story | Morgan Cundiff | 42m |
Claim: only 13% in production of data science projects actually ship. Project built is Phoenix color extractor — a computer vision pipeline that starts with a… | ||
| AI pipelines powered by Jupyter notebooks | Luciano Resende, Alan Chin | 41m |
Shows how multiple Jupyter notebooks become reproducible pipelines by composing them with concrete tools and demos: use of run magic, nbconvert to produce… | ||
| Accelerate large-scale machine learning with NP on MXNet | Hao Jin | 39m |
NP on MXNet claims a drop-in replacement for numpy that keeps numpy APIs while dispatching ops to CPU and GPU; microbenchmarks show about 1.2× on CPU and… | ||
| metadsl: separating API from execution | Saul Shanabrook | 41m |
Claim: metadsl builds a typed, deep-embedded expression graph from normal-looking Python array calls and compiles it to targets like Tensorflow or LLVM; demo… | ||
| Working interactively with large and remote datasets in... | Max Klein | 15m |
Three JupyterLab extensions aim to let the UI handle big remote data rather than forcing full eager downloads: @jupyterlab/hdf5 adds a server extension that… | ||
| Putting Tensorflow Models into Production | Bugra Akyildiz | 01h 18m |
Shows how to go from a trained artifact to a running endpoint: export a saved_model, wrap it in Docker, and register it with TensorFlow Serving so the loader… | ||
| Enhancing Common Natural Language Processing with Cognitive... | Chris Leonard | 39m |
Claims a cognitive layer plus standard NLP improves signal detection by producing a pair of metrics (blue vs red) that map deliberate versus habitual language… | ||
| Building a User Maintainable Visualization Dashboard for... | Dharhas Pothina | 31m |
LSST generates 15 Terabytes per night from a 3.2 gigapixel camera and the team builds a deployable multi‑page visualization dashboard so pipeline faults are… | ||
| Do You Want To Build A Forest? | Thomas J Fan | 31m |
Shows concrete claims and working code: scikit-learn adds histogram gradient boosting (binning avoids sorts, compiled with OpenMP/Cython), matches other fast… | ||
| Productionalizing a Data Science Team | Nicole Carlson | 31m |
Process recommendations come from running two teams that grow from solo to 4 and from 2 to 10 and proving what scales: two-week sprints with daily standups and… | ||
| Visualizing Machine Learning of Units of Measure using PyViz | Jacob Barhak | 33m |
Claims that ClinicalTrials.Gov yields 35,926 trials with results but 24,548 distinct unit strings, so the project builds a pipeline that clusters and maps… | ||
| nothing matches | ||