Bayesian hierarchical modeling
Partial pooling, priors, and full posteriors in PyMC and in Stan, with the diagnostics that say when to stop trusting a fit. NumPyro where the model needs the speed.
Python and R · Modeling trackSports data science, taught live
Learn the work from the people doing the work.
AthlyticZ teaches production data science through sport. Live sessions with named working practitioners, every student in a fully provisioned professional environment in one click. The sport is the hook. The skill is the point.
Upcoming live
Live masterclasses every week, across six tracks.
Every session is taught live by someone who does this work. The recording, the code, and the environment stay with you afterwards.
Next up
What Fills the Seats: Predicting MLB Attendance with Penalized Regression
Fri 28 Aug · 14:00-15:15 EDT / 20:00-21:15 CEST
Greg Matthews, PhD See this week’s sessions
Greg Matthews, PhD See this week’s sessions
Workshop · now booking
Build a no-code analysis board in R, in three hours.
Run with cynkra and taught by David Granjon, using blockr and a board built live on a stage of the Tour de France Femmes. You write your own blocks, against your own data, and leave with the board running.
See the workshop
September 24, then October 1 · three hours · live
For organizations
Put your team on this.
The same practitioners, the same live rooms, run for one group. Organizations come to us when the methods their people need are moving faster than any recorded backlog can keep up with.
Research teams
Keep a group current on methods that move quarterly, taught live by people working in them, rather than working through a course recorded two years ago.
Universities and labs
Run a cohort where every seat opens into the same working environment, so a department is not spending week one administering installs.
Agencies and front offices
Put an analytics group on the tools and methods the people teaching them use in production, with sessions built around the work they actually do.
An institutional license is a flat fee, a number of seats, and provisioned environments for every one of them. The environments run on IDEalyze, our infrastructure platform and a different product of ours. What the fee and the seat count come to is a conversation, not a page.
6Live tracks
23Practitioners teaching
WeeklyLive cadence
30 daysReplay window
How it works
One session, and everything it leaves behind.
The order matters. Each stop exists because the one before it happened.
You turn up
A live room
Taught by someone who does this work, with moderated Q and A, so you ask the question you actually have rather than the one a recording anticipated.
Already set up
The environment
Pre-configured with the packages and system dependencies that session uses, at pinned versions. Nothing to install, no version drift.
It stays yours
You keep everything
Session code, notes and cheat sheets, and a recording with a 30-day replay window. What gets built in the room does not close with it.
Once a month
The Film Room
The Career session takes what the other tracks taught and works it into portfolio and interview material.
What gets built
The skills teams are actually hiring for.
Not a syllabus. These are the artifacts that come up in a technical conversation, in the languages the job descriptions actually name, and every one of them is taught live by someone who builds them.
Deploy and monitor in production
Containerized, pinned, and shipped to cloud: Docker images, CI gates, lockfiles for both languages, and a service that keeps running after the person who built it moves on.
Docker, CI/CD, GCP and AWS · Production trackComputer vision from video
Detection and tracking models over game footage, the labeling workflow behind them, and the extraction pipeline that turns frames into spatiotemporal data you can actually model.
Python, PyTorch · Sports Analytics trackReal-time and streaming data
Live-streaming inputs scored as they arrive, with low-latency paths and the honest question of what a number means when it is still moving.
Python and R · Production trackLLM tools and evaluation
Retrieval pipelines with evaluation harnesses around them, so behavior is measured rather than asserted, and a result can be explained to someone non-technical without hand-waving.
Python and R · LLM Tools trackProduction-grade analytics applications
Full-stack tools people actually open: modular architecture, custom components, R talking to JavaScript over the socket, mobile and offline.
R, JavaScript · App Development trackAgentic AI workflows
Planning work so it can be delegated, running delegated builds, automated browser QA, and validation gates. The loop we run our own work through, taught as a practice rather than a tool demo.
LLM Tools and ProductionEvery one of these is taught live by a working practitioner, in the language the work is actually done in.
Where to start
Four doors, one building.
Everything below runs on the same live rooms and the same practitioners. The difference is what you are trying to get to.
Where you are now
For individuals
Masterclass
Live sessions every week across six tracks, with the recording, the code, and the environment included.
See the schedule
For a career move
Career Programs
A week-by-week game plan: skills, deliverables, mentor sessions, and a capstone you can defend.
See the programs
For organizations
Partnerships
Put a research team, a department, or an agency on current content with dedicated live instruction.
Talk to us
For a build
Consulting
The people who teach this also build it. AI orchestration, modeling, applications.
Book a delivery callEvery live room, the recordings, the code, and the environment, behind one door.
Join the MasterclassStart your path.
One subscription, every live room, and everything the sessions leave behind.
Next live session
What Fills the Seats: Predicting MLB Attendance with Penalized Regression
Fri 28 Aug · 14:00-15:15 EDT / 20:00-21:15 CEST
See this week’s sessions