Terence Z. Liu

You can check out the code here.

You can check out the code here.

You can check out the code here.

Why do you need people management?

turbopuffer is all the rage in vector dbs!

Many people are talking about the power of newer and newer AI agents, and what’s SOTA today is expiring tomorrow.

This post started from researching recent stream processing solutions like RisingWave and Materialize.

There are many aspects of AWS EC2 instances.

There are a lot of analogies between people in a team setting, and computers.

Jotting down the notes.

What’s a “Senior Engineer”?

There aren’t many ways to become better at work as a team.

Note: this doc’s audience is leadership & managers.

Terence shares how Clearview AI’s infra needs evolved and why they chose ScyllaDB after first-principles research.

I presented at the local MadPy meetup group on the new possibilities in the Python ecosystem in light of recent development in the Rust world.

I wanted to write this post after juggling PBs of (non-tabular) data in the past year at work, and having experienced first-hand how amazing the Big Data landscape is morphing into in the mid-2020s.

Started learning Mojo, and watched this deep dive on ownership when it came out: Just summarizing the value and reference semantics across three langauges: C++, Rust, and Mojo: C++: copy: `func(string s)`, caller `func(s)` move: `func(string&& s)`, caller…

You can check out the code here.

You can check out the code here.

You can check out the code here.

I’ve been looking into training some big datasets (500GBs ~ several TBs) of small images with PyTorch.

Updated Oct 7, 2024: use unsafe but correct and faster array content copy to go from 3x to 4x.

In this post, I would like to explore the idea of using embedding vectors to represent code snippets, and compute the cosine similarity scores between a few examples.

I’ve long been curious but hesitant about using embedding vectors generated from pre-trained neural networks with self-supervision.

The stars are aligned.

I originally intended to leave this as a short paragraph, in the middle of a thought piece after learning Rust for a week.

I recently came across an article by Andrej Karpathy, ex-Sr.

This is a quick note on how to use openmp or rather, any multithreading library to divide the underlying data.

For people who fool around in the small field of Approximate Nearest Neighbors (ANN) search, Faiss and hnswlib are two big names.

Not every business needs a Research function, certainly not every startup.

You learned about this in college classes.