home | daniel's blog

👋 Hi, I’m Daniel, a machine learning engineer focused on the intersection of research and product.

Thanks for stopping by. Writings, projects, and contact info below.

Career

If you're here to understand who I am and what I've worked on, my background:

now

Exploring · what's next

Exploring my next adventure in hard tech. Some early-stage investing.

2025

Research · xAI

Frontier language modeling and product development. Shipped the first memory & personalization features.

Acquired X (again) Acquired by SpaceX

2023

Founding Research Scientist, Head of AI · Harvey

Built out several product lines, scaled the AI org from zero to 10+ engineers. Seed stage to post Series C.

2022

Transition Team · X

Helped navigate the transition as Twitter was acquired and converted to X.

Acquired Twitter → X

2019

Senior Machine Learning Scientist · Tesla

Helped build and ship Full Self Driving.

2018

Engineer · DeepScale

Computer Vision for autonomous vehicles.

Acquired by Tesla

Writing

See all posts

I write to clarify my thinking. Publishing forces rigor, encourages better reasoning, and helps me share what I wish I knew years ago.

Topics include:

product development and engineering challenges
psychology of building great products
scaling startups and org design
modern ML research and practical implementation work

I write in a build-in-public style — some posts polished, some exploratory.

Projects

See all projects

As software approaches zero marginal cost, the pipeline from idea → prototype → live on the internet has never been easier. Thinking of the browser as "a way to let other people run my code instantly" makes tinkering fun.

Not everything here is production-grade — some things are just experiments. Recently I've been exploring local-first LLM inference, WebGPU acceleration, and running small models directly in the browser.

Contact

If you want to get in touch with me about something or just to say hi, feel free to send me an email or DM.

X
LinkedIn
[firstname] at [lastname] .sh