Back in Build Mode
Updates on my new L8 SWE/AI role at Axon, the AI Slide Deck, and the production-focused AI Systems course + an open live session tomorrow.
Hey everyone,
I’ve been heads down for the past few weeks.
A few important things have been moving in parallel, and I wanted to take the time to get them right before posting another update here. So instead of turning this into a super long written breakdown, I’ll host an open live session tomorrow and walk through them properly.
Office Hours Tomorrow
I’ll talk through what I’ve been working on, and then open it up for Q&A. I’ll create a Chat thread where you can drop questions ahead of the session, and I’ll use that thread as the backlog.
Here are the 3 major updates on what’s been going on for me.
1. I stepped into an L8 SWE/AI role at Axon, on a mission to Protect Life.
This has taken most of my focus recently.
I’ve been getting up to speed across the systems, processes, culture, AI work, infrastructure, scale, and the places where I can create the most leverage. Getting in touch with multiple stakeholders across teams, from Data to Analytics, AI & AI Infra, Product and Engineering - it’s been a ride, and I’ve learned so much.
At the same time, I’ve joined an amazing team of AI/ML Engineers with tons of real scale experience on their belts, and I’m touching the surface on mentoring and supporting others to improve AI across multiple services and internal workflows.
I’m super passionate about the mission that Axon has, and I’m planning to bring some of those learnings here, and share them with you!
2. The “GPUs and AI” deck has turned into something much bigger.
Back in February, I started working on a few slides about GPUs and AI hardware.
I did not expect the rabbit hole to go this deep.
Since then, I’ve consistently put in ~2h of work every day, Monday through Sunday, to keep reading, digesting, enhancing, understanding and expanding each AI concept.
As of today, my slide deck counts 100+ slides, covering the modern AI stack across hardware, inference, training, foundation models, and the systems around them. Initially, I’ve started with a few topics on key AI Software Frameworks and how to efficiently use GPUs.
But slide by slide, day by day, I’ve got to a state where I already have covered:
Training a Foundational Model End to End
The Data Pipeline Stages
Internals of Transformers/MoEs
From Video Games to AI Compute
Frame Rendering, CUDA Kernels and Vector Rasterizers
and 100+ other topics
Everything structured nicely by chapters, with handmade diagrams/animations.
I’ll share more about the deck soon.
3. My AI Systems Course - Towards a Bootcamp
The course grew into a much larger project than I initially expected.
It is probably the most complete production-focused AI system I’ve built from scratch, and it touches a lot of things I think are still missing from most AI education: starting from PRDs, designing real architecture, thinking of real deployment constraints, real infra decisions, real systems thinking.
More info on this tomorrow.
Drop your questions
I’ll open a Chat thread for questions ahead of the session.
Ask anything you want me to cover: AI systems, GPUs, inference, training, deployment, engineering growth, the course, or the direction of The AI Merge.
I’ll use this thread as the backlog.
Thank you all!
