Scaling an AI Startup in Tier-2 India: Lessons from Ahmedabad

Why Ahmedabad

When Sunbots was founded in 2019, the obvious question was: why not relocate to Bangalore? The Indian startup ecosystem's center of gravity was there — the VC money, the talent pools, the accelerators. The answer was partly personal (this is where I grew up and where I understand the problems), partly strategic (the cost structure of operating in Ahmedabad versus Bangalore is dramatically different), and partly a bet that the problems worth solving in India aren't concentrated in Koramangala.

Six years in, here's what building an AI company in Tier-2 India actually looks like.

The Talent Reality

The honest picture: Ahmedabad has a large and growing engineering talent pool — primarily from GTU, PDEU, and DAIICT, with strong foundational skills. The gap is in applied ML experience. Engineers who have trained and deployed production models are harder to find than engineers who can write Python and have taken a few ML courses.

Our response was to build a learning infrastructure internally. We pair junior engineers with more experienced team members on production projects from week one, run internal reading groups on new research, and create space for engineers to spend 20% of their time on self-directed learning in AI. This has worked reasonably well — several of our strongest ML engineers today joined with foundational skills and developed production ML expertise on the job.

The alternative — competing for experienced ML engineers against Bangalore salaries — is available but expensive. Our approach of developing talent locally has a higher initial time investment but produces engineers who understand our specific domains and are more likely to stay.

The Client Expectation Gap

Global clients — and Indian enterprise clients — have calibrated expectations against Bangalore-based product companies. When you pitch from an Ahmedabad address, there's sometimes an initial credibility gap to overcome.

Our approach: show the work. Our proposals lead with case studies and technical depth, not company pedigree. The SmartON demos, the AI Lawyer platform, the retail theft detection system — these speak more clearly than a Mumbai address. Most of the credibility gap closes within the first technical conversation when it's clear the team understands the problem deeply.

The client gap also presents an opportunity: we often have access to large enterprises in Gujarat's industrial heartland — chemicals, textiles, manufacturing — that Bangalore-centric AI companies aren't actively pursuing. Some of our strongest client relationships are with businesses where we were the first serious AI conversation they'd had.

The Cost Advantage

Operating costs in Ahmedabad are roughly 40–60% of Bangalore for equivalent office space, housing, and non-engineering overhead. This isn't relevant to software margins, but it is relevant to runway and how we price services.

More importantly, engineer salaries are lower for equivalent experience — not because engineers are paid unfairly, but because the cost of living is genuinely lower. An engineer who earns ₹12 lakh in Ahmedabad has significantly more purchasing power than the same engineer earning ₹16 lakh in Bangalore. Our compensation structure reflects this, and it allows us to offer competitive quality-of-life relative to cost.

What the Constraints Teach You

The most useful thing about building in a place with fewer resources is that you build leaner systems. When you can't casually spin up another GPU instance or hire a specialist consultant, you think harder before adding complexity. Our systems are simpler, our code is easier to maintain, and our products are more focused than they might have been if we'd had unlimited resources.

SmartON runs on a ₹8,000 Jetson Nano and an off-the-shelf USB camera. Dhiya NPM runs in a browser tab. These constraints aren't limitations — they're features that make the products accessible to the users and markets we care most about.