Hi, I'm Salman Khandakar.

I have four years of software engineering experience in production, Over that time I've built SaaS products for construction and manufacturing companies, trading and stock market platforms, and compliance-driven systems for healthcare and education — each with its own rules for what "reliable" actually means.

What stays constant is the discipline: systems that don't break under load, and instrumentation that tells you why before a customer has to. Alongside that, I build the growth-focused tools— automations, dashboards, integrations — that help businesses scale once the core platform is solid.

That discipline shapes how I work. is what I bring to AI engineering. I'd rather read logs until the actual cause turns up than guess and redeploy. A lot of my best fixes have been one line long and taken two days to find — and I've made peace with that trade.

I'd rather ship a boring, well-instrumented service than a clever one nobody can debug at 2am. Clever code is a loan against your future self.

Right now I'm putting that same discipline into AI engineering — retrieval, agents, and evaluation — so LLM features arrive as measurable software instead of a demo that worked once on a good day.