Seth Karten

Agents….Continual Harness, PokeAgent, LLM Economist | Research Intern @PrimeIntellect | CS PhD @Princeton | Former CMU Waymo

[Seth Karten @sethkarten]

https://t.co/mTE4Dta508

[Seth Karten @sethkarten]

@ion_barrel Yo

[Seth Karten @sethkarten]

@iamwaynechi @arena @ml_angelopoulos @infwinston @istoica05

[Seth Karten @sethkarten]

Old project but thanks for the feature!! Go check out Continual Harness so we can learn these pokemon harnesses directly from experience

[Tom Dörr @tom_doerr]

Expert-level LLM agent for competitive Pokémon

https://t.co/ZijL3peI7S

[Vincent Weisser @vincentweisser]

Join us at @PrimeIntellect to build the open stack for self-improving agents

Engineering
• MTS – Full Stack Software Engineer — SF/Remote, Full time
• MTS – GPU Infrastructure — SF/Remote, Full time, Hybrid
• MTS – Inference — Remote/SF, Full time, Hybrid
• MTS – Sandbox Platform — SF, Full time, On-site
• MTS – Security — SF, Full time, On-site
• MTS – Training Platform — SF, Full time, On-site

Research
• Research Engineer – Distributed Training — SF/Remote, Full time
• Research Engineer – Reinforcement Learning — SF/Remote, Full time
• Research Engineer – RL Infrastructure — SF/Remote, Full time
• AI Research Resident – Open Source AGI — Remote, Part time

Applied Research
• Evals & Data — SF, Full time, Hybrid
• Forward-Deployed — SF, Full time, On-site
• RL & Agents — SF, Full time

Compute / Finance
• Head of Compute — SF, Full time
• Strategy and Finance Lead, Compute — SF, Full time

Finance / Operations
• Business Operations Lead — Remote, Full time
• Chief of Staff — SF, Full time
• Founder's Associate, Business Operations — SF, Full time

Growth
• Account Executive — SF, Full time
• Head of Enterprise — SF, Full time
• Head of Growth — SF, Full time
• Head of Marketing — SF, Full time
• Revenue Operations Lead, AI Infrastructure — Remote, Full time
• Solutions Architect – AI Infrastructure — SF, Full time
• Technical Account Manager – AI Infrastructure — SF, Full time

Legal
• Head of Legal — SF, Full time

Others
• Internship — SF, Full time
• Open Application for Unconventional Talent — SF/Remote, Full time

[Seth Karten @sethkarten]

@NoahZiems @PrimeIntellect Welcome!!

[Seth Karten @sethkarten]

@alexinexxx Linear switches. Incredible smooth creamy tactile experience and sounds amazing if you are into keyboard asmr. Weighted bar in keyboard.

Only drawdown is it can be quite noisy to coworkers

[Seth Karten @sethkarten]

@code_star The alg isnt just # of followers and the ranking heuristically does seem to capture influence.

Id be interested to see the raw formula though releasing it would def hurt the ecosystem

[Seth Karten @sethkarten]

@TaliaRinger Clearly I am going to Sinnoh

[Seth Karten @sethkarten]

@xeophon prime eval infra is actually great

[Seth Karten @sethkarten]

@iamwaynechi Not clear at all what opus 4.8 did in this video

[Seth Karten @sethkarten]

Yes it can arbitrarily Create, Update, Delete the harness components (system prompt, skills, memories, subagents) through the /refine

I think most harness engineering is about setting up proper feedback loops so the model can observe and refine over time. The concept can be generalized to just about everything dating back to optimal control

[Seth Karten @sethkarten]

Improving your SKILL.md is not a stationary optimization problem.

Continual Harness reveals inherent issues that occur when self-improving skills.

1. Skills can oscillate in performance quality especially when new data reveals a covariate shift.

2. And even when skills strongly dominate basic inputs (think bash/button interface), the model refuses to use them over time

The only way to improve this is to co-train the harness and model to make a true foundation agent

[Seth Karten @sethkarten]

@alexUnder_sky @eliebakouch you can imagine that the rapid adoption of agents for b2c/c2c commerce will create the first living multi-agent system of economic simulacra that I have dubbed the "agent bazaar"

[Seth Karten @sethkarten]

@a1zhang I need my own RLM swarm

[Seth Karten @sethkarten]

@TheAhmadOsman bloat makes it harder for running local open source agents too. You can think of the Refiner from Continual Harness as a generalized compact+optimize for system prompt, memories, skills, and subagents.

Refinde, shorter context makes local agents possible

[Seth Karten @sethkarten]

Go check out more ways we measure reasoning topology and quantitative harness self-improvement performance in the paper and blog post
https://t.co/Wfqw0JU8FL
https://t.co/AxiZz6TQfP

[Seth Karten @sethkarten]

https://t.co/mTE4Dta508

[Seth Karten @sethkarten]

@chijinML Congrats to Dr Ahmed!!