Carlos Galarza carloslfu

Advice on Voice AI, June 2025

My top three pieces of advice for people getting started with voice agents.

Spend time up front understanding why latency and instruction following accuracy drive voice AI tech choices.
You will need to add significant tooling complexity as you go from proof of concept to production. Prepare for that. Especially important: build lightweight evals as early as you can.
The right path is: start with a proven, "best practices" tech stack -> get everything working one piece at a time -> deploy to real-world users and collect data -> then think about optimizing cost/latency/etc.

Setup Guide for `llama.cpp` on Nvidia Jetson Nano 2GB

This is a full account of the steps I ran to get llama.cpp running on the Nvidia Jetson Nano 2GB. It accumulates multiple different fixes and tutorials, whose contributions are referenced at the bottom of this README.

Procedure

At a high level, the procedure to install llama.cpp on a Jetson Nano consists of 3 steps.

Compile the gcc 8.5 compiler from source.

Solving SAT via interaction net superpositions

I've recently been amazed, if not mind-blown, by how a very simple, "one-line" SAT solver on Interaction Nets can outperform brute-force by orders of magnitude by exploiting "superposed booleans" and optimal evaluation of λ-expressions. In this brief note, I'll provide some background for you to understand how this works, and then I'll present a simple code you can run in your own computer to observe and replicate this effect. Note this is a new observation, so I know little about how this algorithm behaves asymptotically, but I find it quite

Some remarks on Large Language Models

Yoav Goldberg, January 2023

Audience: I assume you heard of chatGPT, maybe played with it a little, and was imressed by it (or tried very hard not to be). And that you also heard that it is "a large language model". And maybe that it "solved natural language understanding". Here is a short personal perspective of my thoughts of this (and similar) models, and where we stand with respect to language understanding.

Intro

Around 2014-2017, right within the rise of neural-network based methods for NLP, I was giving a semi-academic-semi-popsci lecture, revolving around the story that achieving perfect language modeling is equivalent to being as intelligent as a human. Somewhere around the same time I was also asked in an academic panel "what would you do if you were given infinite compute and no need to worry about labour costs" to which I cockily responded "I would train a really huge language model, just to show that it doesn't solve everything!". We

So you want to write a sync system for a web app with offline and realtime support? Good luck. You might find the following resources useful.

Overview articles

Database in a browser, a spec (Stepan Parunashvili)

What problem are we trying to solve with a sync system?
The web of tomorrow (Nikita Prokopov)

I heard some points of criticism to how React deals with reactivity and it's focus on "purity". It's interesting because there are really two approaches evolving. There's a mutable + change tracking approach and there's an immutability + referential equality testing approach. It's difficult to mix and match them when you build new features on top. So that's why React has been pushing a bit harder on immutability lately to be able to build on top of it. Both have various tradeoffs but others are doing good research in other areas, so we've decided to focus on this direction and see where it leads us.

I did want to address a few points that I didn't see get enough consideration around the tradeoffs. So here's a small brain dump.

"Compiled output results in smaller apps" - E.g. Svelte apps start smaller but the compiler output is 3-4x larger per component than the equivalent VDOM approach. This is mostly due to the code that is usually shared in the VDOM "VM" needs to be inlined into each component. The tr

Ps: The current setup was done on 01-04-19

Project Dependency Versions at the time 👇

  "react": "^16.7.0",
  "react-dom": "^16.7.0",
  "react-scripts": "2.1.3",
  "typescript": "^3.2.2"
  "tslint": "^5.12.0",
  "tslint-config-prettier": "^1.17.0",

Algebraic Effects in JavaScript part 4 - Implementing Algebraic Effects and Handlers

This is the final part of a series about Algebraic Effects and Handlers.

Part 1 : continuations and control transfer
Part 2 : Capturing continuations with Generators
Part 3 : Delimited continuations
Part 4 : Implementing Algebraic Effects and handlers

So we've come to the core topic. The reality is that we've already covered most of it in the previous parts. Especially, in the third part, where we saw delimited continuations at work.

	<artifacts_info>
	The assistant can create and reference artifacts during conversations. Artifacts are for substantial, self-contained content that users might modify or reuse, displayed in a separate UI window for clarity.

	# Good artifacts are...
	- Substantial content (>15 lines)
	- Content that the user is likely to modify, iterate on, or take ownership of
	- Self-contained, complex content that can be understood on its own, without context from the conversation
	- Content intended for eventual use outside the conversation (e.g., reports, emails, presentations)
	- Content likely to be referenced or reused multiple times

	document.body.innerHTML = 'Paste or drop items onto this page. View results in console.';


	function getPayload(item) {
	const kind = item.kind;
	switch (kind) {
	case 'string': return new Promise(res => item.getAsString(res));
	case 'file': return Promise.resolve(item.getAsFile());
	default: throw new Error('unknown item kind! ' + kind);
	}