We build what actually runs.
A small team with deep experience in production AI systems, product engineering, and the operational foundations that make AI useful in the real world.
RootOps was founded on a simple frustration. Teams were spending months on AI initiatives, building polished demos, and still failing to ship systems anyone could depend on.
The failure mode was rarely the model itself. It was the missing production layer around it: integration, fallback logic, monitoring, edge-case handling, process design, and team adoption.
RootOps exists to close that gap. Not as a consultancy that produces a deck and exits, but as a technical partner that stays close enough to ensure the system runs in production and the owning team can operate it.
We work with startups, growing companies, and institutions across India and globally. The work sits in the space between an AI experiment and a system the business can trust.
Previously VP Engineering at Partex, where he scaled engineering from 40 to 120 and shipped production AI products including MoPilot and Ontosight.ai. Before that, six years at Innoplexus building data intelligence systems in life sciences.
We are looking for people who have shipped AI in production and understand the difference between a demo and a dependable system.
Reach out if that is you →Six things we actually believe.
Not aspirational value statements. These shape how we scope, communicate, and decide what not to take on.
We do not stop at prototypes. If it does not run in a real environment, the work is not done.
We will tell you when AI is the wrong answer, when process is the real problem, or when scope needs to shrink.
Aggressive scoping beats open-ended retainers. The point is shipping something real, not extending the conversation.
Fewer engagements, more senior attention. We are not optimizing for volume.
The system should be maintainable by your team, not dependent on us staying embedded forever.
Capability transfer is part of done. A technically sound system still fails if nobody can use it confidently.
What makes us different from everyone else pitching AI.
Most firms can advise on AI. Fewer have carried production systems through real reliability, scale, and cost pressure.
The edge cases are not only software problems. Prompt design, output validation, observability, and model economics all matter.
The person shaping the architecture is the person staying close to delivery. No bait-and-switch staffing model.
Production AI work sits across product, backend, AI systems, and operations. That needs integrated judgment, not isolated execution.
Let's build something real.
A 30-minute conversation to understand what you want to build or automate. No proposal until the problem and the fit are both clear.