Senior Software Engineer @ DoorDash · 12+ yrs in tech
Real AI engineering.
No hype.
I learn AI engineering in public and publish the part most accounts skip: the builds, the failures, the actual numbers.
$ whoami
senior engineer · distributed systems · 12+ yrs
$ cat current_focus.txt
→ AI agents, token optimization, vector DBs
$ git log --oneline proof-of-work
a3f21c9 prompt caching insight → millions of tokens saved
7e04b1d read claude-mem internals → learned vector DBs
c92d8e4 agent diaries: day 1, shipping not just reading
$ ▊
12+ years
of systems, scale, and production
Millions of tokens
saved with one caching insight
100% real
repos, builds, and failures
What you'll find here
Three kinds of content, one test: does it make you a better AI builder?
Opened Up
Deep dives into real systems. Token caching, vector DBs, repo internals.
Agent Diaries
A numbered log of learning AI agents in public. Wins and failures both.
Engineer → AI Engineer
Your engineering skills carry over. One transferable skill at a time.
From the reels to the real thing
Latest deep dives
How One Prompt Caching Insight Saved Us Millions of Tokens
The full write-up behind the reel. How Claude's prompt caching works, the mistake most teams make, and the simple fix that cut our token costs.
I Learned Vector Databases by Reading claude-mem's Internals
Skip the tutorials. I opened a real open source memory system and traced how it stores and finds memories. Here is the walkthrough, with the code.
Agent Diaries #1: What System Design Taught Me About AI Agents
Day one of learning AI agents in public. Before writing any code, I mapped agent ideas onto systems patterns I already know. The overlap is big.
Newsletter
Get the next deep dive in your inbox.
Free · no spam · unsubscribe anytime
// found me through a reel?
The keyword you commented unlocks the full deep dive.
watch the reel
60 seconds on Instagram
comment the keyword
I DM you the link
read the deep dive
code, numbers, repo