nico.dev
Machine learning engineer

Hello, my name
is Nico.

I am an open source machine learning engineer with focus on WebML at Hugging Face and Google Developer Expert in AI and web technologies, from Switzerland.

Nico Martin

About me

WebML at Hugging Face, Google Developer Expert, and a few years of conference stages behind me. The long version, plus what came before.

Who I am

Talks

AI agents in the browser, K.I.T.T. in JavaScript, robots over Bluetooth, ... Every one with on-device live demos along the way.

See the talks

Latest

Side projects, articles, videos, pictures and podcast appearances. Whatever I am currently building and exploring.

Have a look
Up next

Where you can catch me

Over the past few years, I’ve spoken at conferences and meetups across Europe, meeting brilliant people and happily nerding out about what browsers can do.

All talks and recordingsFind a talk for your event
What's new

Latest things I've done

View all updates
transformersjs-v4
Video
SEP 2026

We shipped 207 WebGPU Kernels for Browser AI

all open source on the Hugging Face Hub, plus @huggingface/kernels, a library that loads, validates, renders and runs them for you. This video walks through how it works: what an op is vs a kernel, why WGSL, and why we ship Jinja templates instead of fixed WGSL files. The browser picks the data types and workgroup sizes your device supports and builds the fastest kernel for that hardware. Then two demos built on top of it: an attention mechanism with the tinyBERT weights in ~20 lines of JavaScript, and a 1M+ cell wave animation driven by a single matmul kernel, capped only by requestAnimationFrame. Finally, Fleet: run the kernel benchmarks on your own GPU. The results help us optimize the kernels for the thousands of devices we can't test ourselves.

Watch on YouTube
Blogpost
SEP 2026

Introducing @huggingface/kernels: 200+ WebGPU Kernels for Local AI

Hugging Face

One of our biggest goals on the WebAI team at Hugging Face is to make browser inference as fast and as user-friendly as possible. Getting there is a multi-layer effort: models need browser-friendly representations, runtimes need to build efficient execution plans, and the individual GPU operations at the bottom of the stack need to make the most of many different devices and browser implementations.

Read the article
Screenshot 2026-08-07 at 08.30.46
Project
AUG 2026

LFM2.5 Edge Agent

nico-martin/LFM2.5-2.6B-WebGPU17

I built a research agent that runs entirely in the browser and puts its full think, plan, and act loop on display. Give it a question and it: - Creates an explicit action plan - Reasons about the next step - Chooses and calls tools - Inspects each result - Updates the plan and repeats It does not stop after one tool call. The agent keeps looping until every step in the plan is complete. Along the way, it can ask clarifying questions, use local context, delegate focused research to an isolated Wikipedia subagent, and collect sources. Once the plan is done, it turns the evidence into a cited response and downloadable research paper. The incredible part is the size: just 2.6B parameters. That makes LFM2.5 seriously fast on-device with WebGPU, while still being capable enough to plan, reason, use tools, and finish multi-step objectives. The model, agent loop, and inference pipeline all run locally. No inference server, no account, and prompts and outputs stay in the browser. Small models are not just autocomplete. They can be agents at the edge.

Launch appSource code