About
About
I'm Ashwin — a data person in New York, working on marketing science at a large financial services company. Most of my day is spent somewhere between modeling, measurement, and the infrastructure that has to exist for either to be trustworthy.
Outside of work, I end up reading a lot more than I write. The list below is a snapshot of what I've been spending time on — less a résumé, more a window into what's on my mind.
Currently thinking about
- The scaling debate. Whether bigger models keep producing qualitatively new capabilities, or we're hitting limits that need a different kind of research. The trendline people and the cognitive-hardware people both have a point.
- Harness engineering. The emerging discipline of shaping how coding agents interact with a codebase — repo-as- system-of-record, agent-legible software, skills and instincts as first-class objects.
- Channel displacement. What happens to marketing, search, affiliate, and the whole distribution stack when agents become the primary interface between users and products. This one hits close to home.
- Membership infrastructure. The quiet shift from transactional commerce to durable relationships, and what kind of substrate makes that work at scale.
- Context as a compounding asset. Write-path data from agents, decision traces, persistent knowledge bases — why being present at the moment of action beats analyzing after the fact.
- The sorting machine in public markets. Why the top decile is eating an ever-larger share of earnings, and how much of that is AI vs. structural.
Recently read / watched
- Situational Awareness — Leopold Aschenbrenner
- AI 2027 — Kokotajlo et al.
- The Scaling Hypothesis — Gwern Branwen
- Dwarkesh Patel's The Scaling Era (oral history)
- Karpathy on why AGI is still a decade away
- Various Stripe / Column / Block writing on financial infrastructure
I keep a private wiki to stitch this stuff together. Some of it may end up in writing, eventually.