TL/IRL: Reinforcement Learning, Foundations to Production
Reinforcement learning has moved from research curiosity to a core technique behind how frontier labs build AI agents.
What started with RLHF has rapidly expanded into a richer post-training stack — RLVR, GRPO, verifiable rewards, reasoning models, and the techniques that defined the field a year ago are already being rewritten. Yet while research keeps accelerating, the gap between a paper result and a production-ready model has widened.
Teams are asking the same questions: Is this a problem RL can actually solve? How do we design a reward that won't get hacked? What's the minimum we need to start? And which tools and platforms are actually worth betting on?
What we'll cover
We'll spend the evening with AI practitioners working at different ends of this stack, covering:
Foundations & frontiers: what's actually changed in the last year
Tools & platforms: the stack shaping how teams build with RL today
Production realities: data design, reward-hacking traps, and lessons that generalize
Plenty of room for questions and tangents throughout. Food 🍕 drinks 🥤 and good company afterward. The hallway conversations are usually the best part.
Agenda
5:30 – 6:00 PM: Welcome & Pizza 🍕
6:00 – 7:15 PM: Speaker Talks 🎤
7:15 – 8:30 PM: Networking 🤝
Speakers
Pashootan Vaezipoor, Head of AI Research, Georgian
Hadi Nekoei, Applied Research Scientist, Georgian, CS PhD Candidate, Mila
Wyatt Kyte, AI Engineer, Adaptive ML
Nima Shahbazi, Head of Data Science, Collective[i]
About Georgian Transferred Learnings
This is the 2nd session of TL/IRL, the in-person AI talk series organized by Transferred Learnings — Georgian's technical community for AI builders. These sessions are carefully designed for AI practitioners and builders who want to explore, learn, and build. True to the spirit of "Applied AI," we share the latest trends and practical insights from builders actually moving the needle.
Who it's for: AI builders (researchers, engineers, founders) who are shipping AI solutions in the wild and have a natural curiosity to stay ahead of where the field is going. The conversations assume you have a foundational understanding of ML and hands-on experience. If you're already building, you'll feel right at home!
✨ We'll also share what we're up to at Georgian and a few open roles across our team and portfolio companies, in case anything catches your eye.