Research × product × code
Gareth Paul Jones
Product Lead, AI — Poe / Quora
I work where model behavior, product judgment, and everyday use meet—then build the prototype, eval, or tool that makes the next decision clearer.
Case studies and public builds
Selected work
Product case studies from public sources, alongside hands-on work on AI evaluation and agents.
App ecosystem · Case study
Poe Apps: from chat to apps
My role in the 0-to-1 launch, the public product history, and a retrospective on creator workflows and app quality.
Read case study (PDF)Model quality · Case study
Zoom AI Companion: routing and quality
My model-product scope, Zoom’s published evaluation results, and a retrospective on routing and launch readiness.
Read case study (PDF)Evaluation
Evals for AI Agents
A practical look at testing agents against the work we actually expect them to do.
Watch the sessionModel behavior
Building a Multi-Signal Reward Function
Working through how several imperfect signals can produce a more useful training objective.
Watch the sessionAgent systems
Managing Coding Agents with Swimlanes
An operating model for keeping parallel coding agents legible, scoped, and moving.
Watch the sessionLive Streams
89 videosSwipe or use the arrow keys to scroll. Tab to each video.
Writing
Using Time Blocking or Calendar Purging to Supercharge your Day
Maximize Productivity and Organize Your Schedule with Effective Time Management Techniques
TensorFlow Serving 101 - How to Deploy Your TensorFlow Model
An Introduction to Deploying and Managing TensorFlow Models in Production Environments
Writing Custom Text-To-Speech with your own voice
Embrace the Power of Voice - A Step-by-Step Guide to Custom Text-to-Speech
