Frontier AI01
LLM agents, retrieval, eval infrastructure, and the messy operational layer that turns prototypes into businesses.
01Who we are
I founded this studio
to keep learning out loud.
I am Sal Anvarov — engineer, founder, and the person behind DEV BY SAL LTD.. I started this studio because the problems I find most interesting do not live inside a single company or a single industry. They move.
02The short version
A studio that takes the work I would have wanted to take anyway.
Open to new engagements
I founded DEV BY SAL LTD. because the kind of work that interests me does not fit cleanly inside any one company. One month it is a retrieval-grounded AI layer for a wealth-management product. The next it is real-time substitution intelligence for a national grocer. The month after that, an auditor portal that unifies three legacy compliance systems behind a single SSO.
The variety is not a side effect of running an agency. It is the reason I run an agency. I am fascinated by learning, allergic to repetition, and happiest at the edge of what I do not yet know how to build. A studio is the structure that lets me show up to a new problem every few weeks — and the only honest way I have found to keep getting better at my craft.
The work we ship is consequential. It runs in production at regulated enterprises, at grocers handling millions of orders, at AI infrastructure companies, at non-profit edtech serving learners in twelve languages. Every engagement is run by senior engineers with no project managers in between, and we say no to the work we are not the right shop for. There is a longer version of how we actually do that on the studio about page.
If you want to know what I think about, what I am building outside this studio, or what I have written that did not fit here, the fastest route is sal-anvarov.com.
— Sal
03Where the attention goes
Four edges I keep
walking toward.
I work where the problem is technically ambitious and the customer is real. These are the frontiers I currently care most about.
Frontier software02
Distributed systems, internal platforms, developer tooling — anything that compresses a team’s ten weeks into one.
Frontier industries03
I take work from sectors I have not worked in yet on purpose. Insurance one month, EdTech the next. The variety is the point.
Frontier conversations04
I read a lot, I write some of what I read, and I learn fastest in rooms with people more interesting than me.
04How I think
Four operating principles,
said plainly.
01
Curiosity is a competence, not a personality trait.
The engineers who keep getting better are the ones who treat learning as the actual job, not a side effect of the work.
02
Variety beats specialization, past a point.
A new industry every quarter teaches me more than a tenth year inside the same one would. Different surfaces, same underlying craft.
03
Hard problems are the only ones worth taking.
If a problem can be solved by a generic SaaS, it should be. I take the ones that can’t.
04
The work speaks for the studio.
No leaderboards, no glossy decks. The case studies are the pitch. The papers are the credentials.
Most of what I do shows up on this site as case studies, papers, and writing. The rest — the half-built experiments, the side projects, the longer-form essays I have not yet pulled into a piece — lives on my personal site.
If you are evaluating whether to work with the studio, the case studies and papers here will give you the clearest picture. If you want to know the person behind the work, that is what sal-anvarov.com is for.
Working on something genuinely hard?
30 minutes. Honest answer. If we are not the right shop, we usually know who is.