AI Engineer

I build AI systems that hold up in production.

I am Raj Gajjar, an ai engineer in Ahmedabad, India. I work on retrieval, agents, and the API layer underneath them, mostly in Python and FastAPI.

I'm an AI engineer in Ahmedabad, and I'm most curious about software that acts on its own: agents that plan before they touch anything, voice assistants you can actually hold a conversation with, and tools that turn plain language into real work. Outside of that, I like digging into how the things we take for granted work under the hood, like TLS, and building interfaces with a bit of motion in them, like the plane you've been following down this page.

How I work

What working with me looks like

  1. First I learn what a right answer looks like to you.

    Not the roadmap, the actual work: your data, your edge cases, the questions your users type at 2am. AI projects fail on that definition far more often than they fail on the model.

  2. You see the thing running, not a status report.

    Every week there is something you can open and try against your own data. Nothing important gets discovered in the last sprint, when changing direction costs the most.

  3. You get to see how often it is wrong.

    Real cases from your world, the answers you would accept, a score that reruns on every change. Nothing ships here because it demoed well once on a Tuesday.

  4. I work inside your setup, not around it.

    Your repo, your review cadence, your pace. A two person startup and a team of forty need very different amounts of process, and the handoff should run on yours without me in the room.

Have something worth building?

I am open to AI engineering and backend roles, and to contract work on retrieval systems, agents, and Python APIs. The fastest way to reach me is email.