Skip to content
Pareto Inference

Pareto Inference serves open and custom models through one OpenAI-compatible API, tuned to each customer's traffic. We started as Precursor Labs, a research company working on scheduling and memory management for inference serving, and now go by one name.

Team

A small, in-person team in NYC, working across accelerators, architectures and every layer of the stack.

  • Jakub Janiak

    CEO

    M&A investment banking at Centerview Partners. Runs partnerships and compute.

  • Niyant Narang

    CTO

    Reinforcement learning at Northrop Grumman. Graduate AI at Stanford. Leads engineering.

  • Dylan Vu

    COO

    Chemical engineering research at Cornell, Harvard and L3Harris. NSF fellow; left a Berkeley PhD. Runs operations.

  • Alex Janiak

    Chief Science Officer

    Low-level systems and caching on AWS Lambda. Computer science, math and philosophy at Duke. Leads research.

Backers

A $4M seed round led by BCAP, plus NVIDIA's program for AI startups.