Learning in public
Turning what I learn into things you can use.
I'm Keyvan. This is my working library: structured paths through subjects I'm learning, practical playbooks from work I've shipped, and short updates on what changed.
On the desk now
rebuilding how a large multi-repo team works with coding agents — spec-driven, small PRs, enforcement in hooks rather than prose.
Go deep
Structured learning
Do the work
Practical field guides
Stay current
Signal over noise
Now
Learning by making it useful.
Open loops
What I'm working through
- 01
Agentic delivery on a real multi-repo product
Turning tribal knowledge into artifacts agents can load, and standards into checks a runtime enforces. Most of what I've learned is written up rather than theorised.
Read the playbook - 02
Evaluation over vibes
Getting past 'it seems better' — building eval harnesses that catch regressions before users do.
Chapter: eval-driven development - 03
Leading without authority
The part of the job that isn't technical: driving adoption across teams and business units that don't report to you.
Chapter: the operating model
The library
Choose how deep you want to go
The useful bits from this week.
A short note on what changed, why it matters, and where it connects.
All updates- 01
Building effective agents
Anthropic
- 02
Your AI Product Needs Evals
Hamel Husain
Nothing here is finished. That's the point.
Written in the open · improved in public