AI/ML Research Engineer at YC W26 enterprise AI startup

You will join a 4-person founding team to build high-speed AI agents that automate legacy enterprise workflows. Moving beyond slow screenshot-to-VLM loops, you will develop agents that pre-train on interface navigation to achieve 5x faster execution. This role blends cutting-edge research in agentic orchestration with rapid production deployment.

• Research and implement novel agentic architectures for GUI automation using multi-agent coordination, memory, and context management.
• Build and evaluate reasoning pipelines—including chain-of-thought and reflexion loops—that maintain reliability under distribution shifts in enterprise environments.
• Develop interface pre-training methods and VLM-based screen understanding to enable deterministic execution and self-healing for automated enterprise agents.

• Early-career researcher (0-4 years) with a Master's or PhD in CS/AI from a top-tier program or a track record at a premier research lab.
• Strong engineering skills in Python, PyTorch, and agentic frameworks like LangGraph or AutoGen, with the ability to move from paper to prototype rapidly.
• Deep curiosity for computer-use agents and GUI understanding, evidenced by top-tier publications (NeurIPS, ICLR, CVPR) or significant production-grade AI projects.

• Join a Y Combinator W26 company at the ground floor, working directly with founders on the core technology that defines the product's intelligence.
• Solve a massive enterprise bottleneck by building deterministic, self-healing agents that operate complex legacy software without APIs or structured data interfaces.
• High-impact environment where your research in reasoning models and vision-language architectures is shipped to production for real enterprise customers immediately.