PyData Virginia 2025
2025-04-18..2025-04-19
Charlottesville, United States
who it's for
data scientists, machine learning engineers, data engineers, python developers, ml researchers, data analysts, mlops engineers
topics
variational inference with ADVI and Pathfinder; agentic AI multi-agent pipelines; DuckDB in-process analytical SQL; Apache Arrow zero-copy IPC; prompt engineering and chain-of-thought; retrieval augmented generation strategies; pandas to PySpark migration; GPU-accelerated RAPIDS workflows; multimodal model evaluation and versioning; image classification with scikit-image and PyTorch; bayesian changepoint and risk analysis; graph analytics with NetworkX and cuGraph
| title | author | length |
|---|---|---|
| A Beginner's Guide to Variational Inference | Chris Fonnesbeck | 01h 29m |
Tutorial shows concrete claims and timings: PyMC runs ADVI and the newer Pathfinder and compares them against MCMC NUTS on real models. Example: Gaussian… | ||
| Build Your Own Data Science AI Agents | Krishnan, Liu, Puri, Rojas | 01h 29m |
Claim: a runnable multi-agent pipeline that reads diabetes.csv, does EDA, trains a predictive model and auto-generates an executive PPT. Tools actually used… | ||
| Data Wrangling with DuckDB | Will Angel | 34m |
DuckDB is described as an MIT-licensed embedded analytical SQL database that runs in-process, uses Arrow to shuttle data to and from Pandas and Polars, and… | ||
| Practical Applications of Apache Arrow | Will Ayd, Matthew Topol | 35m |
Apache Arrow specifies a columnar in‑memory layout that is identical on the wire so systems achieve zero-copy IPC across C, Python, Go, Java and more and avoid… | ||
| Mastering LLMs: From Prompt Engineering to Agentic AI | John Berryman | 01h 39m |
Covers concrete workflows: prompt tricks like few‑shot, chain‑of‑thought and ReAct for grounding; RAG implemented as tool calls with both vector versus lexical… | ||
| From Pandas to PySpark | Cynthia Ukawu | 01h 13m |
Hands-on roadmap for taking pandas code to PySpark using an anime recommendation dataset in Colab, showing concrete commands to install PySpark, create a… | ||
| Practical Multi Armed Bandits | Benjamin Bengfort | 30m |
Concrete takeaways: implemented systems combine epsilon-greedy with Thompson sampling, compute updates in NumPy and SciPy, and store minimal state for fast… | ||
| Tutorial on Image Classification using Scikit-Image, Scikit-learn, and PyTorch | Matt Litz | 01h 15m |
Pipeline that actually runs: scikit-image for segmentation (adaptive histogram, multi-Otsu, remove_small_objects, morphological ops and a WA segmentation… | ||
| RAPIDS: GPU-Accelerated Data Science for PyData Users | Naty Clementi, Mike McCarty | 01h 36m |
Claim: RAPIDS makes end-to-end PyData pipelines run on NVIDIA GPUs with cuDF (GPU-accelerated pandas) and cuML / QML (GPU scikit-learn), showing real examples… | ||
| Responsible AI with SciPy | Andrea Hobby | 53m |
Shows concrete code and results: bias detection using scipy.stats on the NHANES CKD dataset, a RandomForest model with 0.94 accuracy but heavy class imbalance… | ||
| Blazing the AI Trail: Using LangGraph to Conquer the Oregon Trail | Robert Shelton | 01h 27m |
Hands-on build of a working agent workflow that actually binds Python LangGraph tools, implements vector retrieval against a Redis vector store, and adds… | ||
| KEYNOTE: Rajkumar Venkatesan- Building AI-First Organizations & Opening Notes | Rajkumar Venkatesan | 58m |
Claims and concrete lessons: data remains a strategic asset, foundational models cost hundreds of millions while fine‑tuning runs to thousands, and a five‑step… | ||
| Using Python to Unlock Insights from OpenStreetMap Data at Scale | Cory Eicher | 25m |
Python code automates downloading OpenStreetMap power and telecom data via Overpass API, converts OSM JSON to GeoJSON, then loads into an ESRI geodatabase… | ||
| Building a Robust Evaluation Framework for GenAI Productivity Tools | MacKenzye Leroy | 30m |
S&P Global describes an enterprise evaluation framework that measures retrieval and generation separately for a RAG tool called MI sales assistant, uses MLflow… | ||
| Zero Code Change GPU-Powered Graph Analytics with NetworkX and cuGraph | Ralph Liu | 32m |
NetworkX now dispatches graph algorithms to GPU backends so existing code runs much faster with zero code change. The demo shows enabling dispatching via an… | ||
| Making the most of test-time compute in LLMs | Suhas Pai | 31m |
Claims that extra inference time lifts accuracy using parallel scaling (best-of-n ensembles) and sequential scaling (self‑verification and backtracking), with… | ||
| Bayesian Risk Analysis For Large Multi-Modal Data | Sihang Jiang | 31m |
Claims concrete results: wastewater SEIR models estimate transmission parameters using viral loads and reported cases; 47,000 long COVID cases are matched… | ||
| Unlock Information from Tables, Images and Complex Docs | Krishna Rekapalli | 01h 30m |
Claims that Docling extracts structured data from tables, images and complex multi‑column layouts and produces a Docling document JSON with hierarchical… | ||
| Introduction to Wikidata | Robin Isadora Brown, Lane Rasberry | 01h 13m |
Demo shows running SPARQL‑style queries (referred to as Sparkle) from Jupyter notebooks using a SPARQL wrapper or kernel, pandas and IPython display to pull… | ||
| Versioning Multimodal Data: Metadata & Beyond | Dmitry Petrov | 28m |
The DVC team argues that multimodal needs a metadata-first versioning approach because files are heavy (megabytes to gigabytes) and datasets scale to thousands… | ||
| How data science shortened the COVID-19 pandemic by 2 months | Greg Michaelson | 29m |
Claims that data-driven geospatial and predictive models cut U.S. vaccine trial time by 6–8 weeks, supported by concrete numbers about enrollment speed and… | ||
| Using Changepoint and Bayesian Analysis to Drive Safety Improvements in Mining | Mauricio Mathey | 28m |
Claims and results are concrete: changepoint detection with PELT with Poisson changepoint finds few shifts, Bayesian MCMC gives confidence about those shifts… | ||
| Data Viz in Pyton as a Tool to Study HIV Health Disparities | Kimberly Deas | 01h 07m |
Python code uses Matplotlib, Seaborn, Plotly to turn Texas HIV records plus county health rankings into maps, bar charts and a correlation heatmap that finds… | ||
| Celebrating Neurodiversity Through Aesthetic Data Visualization | Siwen Liao | 27m |
Project claims that cortical-volume patterns separate by sex and by ASD status and that an interaction may mark female-specific differences. Analysis uses… | ||
| Maximizing Multimodal: Exploring the search frontier of text-to-image models | Nathan Day | 26m |
Claims concrete: Eezy swaps CNNs for CLIP/OpenCLIP embeddings to power reverse image search (launched May 2023) and reports visibly better results so obvious… | ||
| Exploring Eviction Trends in Virginia | Michele Claibourn, Samantha Toet | 29m |
Uses Civil Court Data Initiative records and a quarterly searchable dashboard called Virginia Evictors Catalog, plus Poisson and mixed effects logistic… | ||
| AI Ready Data | Hamish Brookeman, Alec Gosse | 26m |
Claims a public open metadata marketplace that assigns a unique URL to every column in customer‑facing tables, exposes machine‑readable JSON tied to ontologies… | ||
| nothing matches | ||