​Welcome to our second meetup dedicated to AI researchers at University of California, Berkeley!

​Agenda

​​​​​🕒 5:30 pm – 6:00 pm
Welcome Gathering amd Opening Remarks

​🎤 Marijan Cipcic, Principal Events Manager at Daytona

​🕒 6:00 pm – 6:15 pm
Talk "Today's Agents Don't Live In Episodes"

​🎤 Lovre Pesut, AI Engineer at Daytona

​​​​​​​Outline:

​The 'episode' (short, stateless, resettable) has been RL's foundational abstraction since ATARI. It underpins the Gym API, GRPO, PPO, and the conventional sandbox lifecycle. Today's agents no longer fit it. Tasks span for days; the env state at hour 18 of an agent session with warm caches, installed deps, live processes, open sockets, dirty git tree, is worth hours of wall clock to reproduce.

Three things are scaling simultaneously. Rollout horizon: seconds -> days. Env state: disposable between episodes -> first-class learning substrate. Branching: absent in modern LLM-RL -> speculative fork trees. Each stresses the inherited toolkit in a different way, and all three have been gated on the same missing primitives: VMs you can fork cheaply, pause without killing processes, snapshot mid-run, and resume hours later.

This talk walks through what opens up when those primitives become available. Live demo of long-horizon sessionful rollouts, mid-trajectory forking, and cross-calendar-time training. The research questions that follow (long-horizon benchmarks, speculative RL algorithms, event-driven training, to name a few) are where the next wave of agent RL gets built.

​🕒 6:15 pm – 6:30 pm
Talk "FrontierCS 2: A Benchmark for Computer Science R&D"

​🎤 Hanchen Li, PhD researcher at UC Berkeley’s Sky Computing Lab

​​​​​​​Outline:

​Can AI agents make meaningful contributions to frontier computer science research? FrontierCS 2 introduces 48 open-ended research tasks across 13 domains from MLSys to security. Each task starts from a published paper and its code, removes the key contribution, and challenges agents to rediscover it through iterative experimentation in a real research environment. We evaluate 12 frontier agents, with the strongest achieving a 58% pass rate, while 16 tasks remain unsolved by any agent.

​🕒 6:30 pm – 6:45 pm
Talk "The Next Hill: What Evals Need to Measure Next"

​🎤 Xiangyi Li, Founder at BenchFlow. Author of SkillsBench

​​​​​​​​Outline:

​Benchmarks are saturating fast, especially in domains with stable fundamentals (finance, law) or easily verified results (math, coding). What stays hard is human interaction, company context, and fields like AI and science where new knowledge keeps appearing. This talk traces how eval formats evolved, from MMLU to SWE-bench and PaperBench. It then argues that the next wave of evals will cover science and AI research, and will measure alignment and safety, agent behavior (over-eager, under-eager, reckless), and multi-agent and physical-world tasks. Examples come from BenchFlow's BenchGuard and FrontierPhysics.

​​​​​​​​​​🕒 6:45 pm - 8:00 pm

​​Networking

​With food and beverages

​​About event

​An engaging meetup designed for AI researchers to connect, share ideas, and explore the latest advancements in artificial intelligence. The event features informal networking, short talks, and discussions on current research trends, fostering collaboration and knowledge exchange within the AI community.

​Venue: Stephens Lounge is located on 3/F, Martin Luther King Jr. Building (UC Berkeley Student Union) at 2495 Bancroft Way, Berkeley, CA 94720