Skip to content

Nicholas Barbier

Working on inference and systems.

About

Mt. Lafayette looking towards Little Haystack, Franconia Ridge
Mt. Lafayette looking towards Little Haystack, Franconia Ridge, NH © Feb 2023
A Tribute to Jascha Heifetz, photomontage by James Arkatov and David Hockney
A Tribute to Jascha Heifetz, James Arkatov in collaboration with David Hockney © 1991
Mulholland Drive: The Road to the Studio by David Hockney
Mulholland Drive: The Road to the Studio, David Hockney © 1980

I was born in the US, raised in Venezuela until I was five, then lived in Boston for thirteen years. I studied at Stanford and lived in the Bay Area for five and a half years. I’m now happily back in New England.

In college I majored in Symbolic Systems with a concentration in mathematical logic. My master’s was in AI, though I also studied operating systems, networking, databases, and programming language theory.

I work across statistical modeling, data engineering, distributed training infrastructure, and inference performance optimization. Most of my work is in Python and, more recently, Rust. I like making systems run quickly, simply, and resource efficiently. The best software doesn’t get in your way.

Right now I’m at Amira Learning, working on speech recognition and inference that listens to millions of students. I also publish personal open-source work under electricapp.

Outside of work: playing classical violin, skiing and hiking on the east coast, tinkering, and keeping an eye on my Siamese cat.

All words are my own.

Experience

Education

  • 2018–19Stanford University, MS Computer Science
  • 2014–18Stanford University, BS Symbolic Systems

Open Source

  • aethergraph — Billion-scale graph sampling for GNN training.
  • ane-bridge-rs — Rust bindings for the Apple Neural Engine.
  • ccaudit — Fast Claude Code log viewer for tokens, cost, and history.
  • hasp — Workflow pin scanner and sandboxed GitHub Actions runner.
  • pgbattery — Raft HA manager and TCP gateway for PostgreSQL.
  • power-monitor — Apple Silicon power monitoring via IOKit, with TUI and menu bar app.
  • sme-gemm — SME GEMM kernels for Apple M4+.

Teaching

Blog

Coming soon.

Photography

Coming soon.

Contact