I'm Merl Chandana, an AI researcher and software engineer passionate about machine learning, software engineering, and the intersection of technology and entrepreneurship.

My work focuses on building and understanding AI systems, with particular interest in how we can develop more capable, useful, and reliable machine learning models. I enjoy tackling problems that combine technical depth with real-world applications, and I'm always curious about new approaches and emerging techniques in the field.

Work

I've worked on various projects spanning research, engineering, and product development. My experience includes working with large language models, building scalable systems, and exploring novel applications of AI technology.

[Update this section with your specific work history, roles, companies, and key accomplishments.]

Outside Work

When I'm not coding or researching, I enjoy exploring new ideas, reading, and working on side projects. I believe in maintaining a balance between deep technical work and other pursuits that keep me curious and energized. I'm also interested in startups and entrepreneurship, and I enjoy thinking about how technology can solve real problems.

About this Site

This website is built with FastHTML, a Python web framework for fast, scalable web applications with minimal, compact code. I was inspired to build it after reading the book Hypermedia Systems, which describes a simpler approach to web development based on the fundamental ideas and technologies that empowered the early web. You can view the full source code for this site on GitHub.

---

Find my full CV here.

Now

Last updated: November 27, 2025

This is a "Now page"—it details what I'm focused on right now, at this point in my life.

I recently left my role at Agemo after an amazing couple of years. I learned so much, had the freedom to work on cool projects and the resources to train LLMs and experiment with the latest techniques, but with so much happening in AI right now, I started to get an itch to try something new.

Rather than jumping straight into the next thing, I decided to take a couple of months to invest in myself—working on personal projects and diving deep into areas I'd been meaning to explore. Purely for fun, and because I was blown away by the book Hypermedia Systems, I've been teaching myself web development. Starting very much at zero ("Claude, please explain what CSS is"), I've built this website entirely from scratch, in a web framework LLMs know very little about.

I've also used this opportunity to take part in the following two courses:

• Scratch to Scale: Large-Scale Training in the Modern World by Zach Mueller from Hugging Face, on the art and science of parallelising LLM training across hundreds of GPUs. I spent a long time at Agemo figuring this out for myself, building up my training recipes, DeepSpeed configs and SLURM scripts from scattered tutorials, Discord discussions and random gists. I was amazed at how hard it was to find high-quality guidance online. So when I saw this course advertised and the expert panel of guest lecturers (from Meta, Anyscale, Unsloth, Axolotl, Prime Intellect), I knew it would be worth jumping on.

• How to Solve it with Code from Answer.AI. Ever since I stumbled across nbdev in 2022, I've been a huge fan of Jeremy Howard: he's contrarian, impervious to fads and seemingly always ahead of the curve. In this course, Jeremy, Johno Whitaker and guests teach the extremely considered and intentional approach that they take when working with AI, be it for solving problems, building new products, writing prose or learning new material. It's a really fun course packed with invaluable insights about everything from web scraping to web development, sys admin to sasquatches.

I'm now turning my attention to new opportunities. If you're working on something interesting, I'd love to hear from you—LinkedIn, X and email are in the footer below.