Full-Stack & AI Engineer
I build software the way problems actually demand it - backend and system design, real product sense, and AI that people can rely on. I'm not defined by a stack; I learn what a problem needs and ship it.
The engineer teams call when it has to work.
Less a list of tools, more a way of working. Here's what people actually get when they bring me in.
I think at the system level - data models, services, and the trade-offs that decide whether a product stays fast and calm under real load.
I build with models the way engineers should: retrieval, vector search, and agentic systems (including MCP) shaped into features people trust, not demos.
I care how a thing feels to use. I design the flow around the person, then build the backend that makes it honest.
I'm not defined by a fixed stack. I learn exactly what a problem needs and bring it to bear - which is why teams call me when it matters.
Where I've worked
I don't measure myself by a list of side projects. What matters is the work I've done inside real teams - the role I played, and what shipped because of it.
2024 — Present
Enterprise SaaS Platform
Owned features end to end across a high-traffic product: data modeling and services on the backend, considered UX on the frontend, and shipped AI capabilities - retrieval and agentic flows - that people actually depend on in daily work.
2023 — 2024
Mid-scale Product Company
Designed the data and service layer for a growing platform - the kind of foundational work that decides whether a product stays fast under load. Stepped in wherever a problem was hardest, from schema design to delivery.
2022 — 2023
Product Startup
Wore every hat a small team needs: built product surfaces, wired up infrastructure, and learned whatever each problem demanded to keep shipping. Reliability under ambiguity became the habit here.
Some of my writing
Notes on the things I think about - AI you can trust, systems that scale, and staying adaptable in a field that never sits still.
RAG, vector search, and agents are easy to demo and hard to trust. Here's how I make them dependable.
Why I don't define myself by a language or framework - and why that makes me more useful, not less.
Scale is decided at the data layer long before traffic arrives. A few principles I keep coming back to.
Let's build something
Tell me what you're building or where you're stuck. I read every message and reply personally - usually within a day.
Prefer email? hello@ishanto.com
© 2026 Shanto Islam