Generalintelligencecompany
Agents Research Lead
9mo ago
400000 –600000 USD / yearUSALeadpythondistributed systemsmachine learningevaluationdataset curationplanningprompting
Lead the applied AI team developing autonomous agents, overseeing experiment design, evaluation, and production deployment.
Responsibilities
- At General Intelligence Company, we’re building highly-capable autonomous agents for startups. Our goal is to enable the one-person, one-billion-dollar company. Our agents don’t just automate workflows - they replace them. We're looking for a researcher to lead our applied AI team to work towards our goal of fully autonomous businesses.
Other
- Design and run experiments across model selection, prompting, tool-use, memory, planning, and multi-agent coordination
- Lead evaluations: build datasets, success criteria, and continuous benchmarks for real workflows
- Push performance: reduce tail latency and failure modes; improve determinism, throughput, and cost per successful run
- Partner with eng/product to ship weekly
- Mentor a small, senior team; set standards for experiment design, code quality, and documentation
- Push state-of-the-art results on custom agent systems
- Pursue advanced memory research for multi agent systems
- 5–8+ years in applied ML/AI, ML systems, or research engineering; high-ownership startup experience preferred
- Deep experience with LLMs and agentic systems: prompting, function/tool calling, planning, retrieval/memory, evals
- Track record moving paper → prototype → production (Python; comfort with distributed systems a plus)
- Published papers in AI/ML that are directly applicable to modern agentic systems
- Built reliable evaluation harnesses and curated datasets for multi-step tasks
- Shipped improvements that moved core business metrics (success rate, latency, unit economics)
- Bonus: orchestration frameworks, streaming/real-time state sync, safety/guardrails, reinforcement/online eval loops
- In-person in New York; high autonomy; ship fast; iterate daily
- Small team, zero politics; you own outcomes end-to-end
- Heavy use of modern AI tooling (Cursor, structured outputs, eval pipelines)
- High performance team working on the most interesting problems in agents.
- No bureaucracy. We hire smart, high-agency people and let them work.
- Real ownership. You won’t just push tickets—you’ll shape the product.
- Small, elite team. Work directly with founders and have a major impact.
- Craft matters. Speed is key, but quality is never sacrificed.
- In person, in New York City.