Our team

Coding has changed more in the last two years than in the twenty before it. The languages, the tools, and the type systems we all rely on were designed for a world where humans did the writing. We're a small team rebuilding those foundations for a world where AI does most of it.

The Boundary team

People building BAML

  • Co-founder & CEO

    Vaibhav Gupta

    Vaibhav previously built the on-device depth and face-unlock systems for Google’s Pixel 4, and worked on Microsoft’s HoloLens.

  • Co-founder & CTO

    Aaron Villalpando

    Aaron previously spent seven years at Amazon, scaling EC2’s internal monitoring and building live-streaming features for Prime Video and Twitch.

  • Engineering

    Sam Lijin

    Sam previously worked on user identity and Cloud Firestore at Google, then built developer tooling at Trunk, after studying CS at Vanderbilt.

  • Engineering

    Antonio Sarosi

    Antonio has written an ACID-compliant database, an Nginx-style reverse proxy, and a memory allocator from scratch in Rust, and teaches it all to 180K+ subscribers on YouTube.

  • Engineering

    Paulo Rossi

    Paulo is a former Y Combinator founder who grew his startup past $2M in ARR, and was an early BAML user before he joined.

  • Engineering

    Kai Orita

    Kai built ere, a Rust crate that compiles regular expressions at build time into efficient, type-checked code, and a VS Code extension with over a million downloads, after an MS in computer science.

  • Engineering

    Avery Townsend

    Avery revived the defunct LEGO Universe MMO with a custom C++ server emulator in high school, then wrote a full wgpu rendering backend from scratch and contributed to the Slint compiler while finishing a degree at UPenn.

  • Engineering

    Dhilan Shah

    Dhilan is a computer science intern from UT Austin who built agent-tries-baml, the system BAML uses to measure and improve how well AI agents write it.

Work on the hard parts.

We're looking for people who can move between product and systems, write clearly, debug patiently, and care about making AI software less fragile.