We place senior AI engineers inside your team, on your tools, in your stand-ups. Not a consultancy. Not a pilot. Real people, joining your roadmap.
Sachith at his Desk
Onboarding day, April 2026
“Most teams I've worked with treat AI as a feature you bolt on later. At Stekz, the LLM layer is wired into the platform before the first product is built. It changes how you design everything.”
Chandra builds digital products on Stekz's proprietary platform, MyStekz — each one architected around Domain-Driven Design, meaning the product logic maps directly to how the client's business actually operates, not to generic technical assumptions.
Every product built on MyStekz connects automatically to BEP, Stekz's own LLM layer. That means conversational AI isn't a feature request that comes later — it's available from the first release, wherever it creates value. The primary development environment is Claude Code.
The result is a stack where business alignment and AI capability aren't in tension. Stakeholders stay close to the product because the domain model speaks their language. And AI is structural — not a layer on top, but part of how the product thinks from day one.
Beyond his work at Stekz, Chandra leads Gapstars' Data & AI Guild — the internal community keeping the wider engineering team current as the field moves. If something significant shifts in how AI is being built, Chandra is usually the person who's already working with it.
“We don't start with ‘what can AI do?’ We start with the manual, repetitive work our engineers are doing, and we automate it. The goal isn't to replace humans. It's to free them up for higher-level problem solving. When AI handles the boilerplate, engineering velocity goes up exponentially.”
Sachith builds generative AI workflows inside Okra.ai's internal platforms, automating tasks that engineers previously did by hand. If there's an established workflow for it, he looks for a way to let an agent do it.
His work includes deploying Claude models to write test coverage, and automating API documentation generation based on codebase changes. These aren't speculative projects — they're tools in active daily use by Okra.ai's engineering teams, saving hundreds of hours a month.
Sachith is a founding member of Okra.ai's dedicated internal AI squad. If an internal process can be made faster with a model, his team is probably already prototyping it.
“The problem with most enterprise AI is that it's top-down. Leaders read about a tool and try to force it into a workflow. True impact comes from the bottom up.”
“When you look at the raw data, the inefficiencies become obvious. You see where people are spending hours on tasks a model could do in seconds. That's where you start. You build small, targeted solutions that solve real problems, and you scale from there. It's not about finding a use case for AI; it's about finding a use case for efficiency.”
Sachintha has spent the last year working across the business at Brompton, identifying where generative AI can drive tangible value. He builds prototypes, tests them with users, and scales the successful ones into production.
Most of what you need is already here. Senior AI engineers, hand-picked for your stack, embedded in your team within weeks. If you're not sure where to start, that's what the second layer is for.
Senior AI, ML and MLOps engineers placed inside your team. They join your stand-ups, ship to your repo, and report to your tech lead. We handle the people: you run the roadmap.
For partners who want to build genuine AI capability inside their teams — not just adopt the tools, but develop the judgement to use them well.
We don't only train engineers in AI. We back the ones already curious about it. Case in point: last
month our teams built multi-agent systems in 24 hours.
One of those builds – a QA automation workflow from the third-place team is shipping into
production at Stekz this quarter.
Real multi-agent architecture, not single-prompt
demos.
Engineering the problem in front of them — not
pattern-matching to last week's tutorial.
Risk scoring, exports, mocked integrations. The brief
was the floor, not the ceiling.
of our engineers say they're ready to apply this in client work
tomorrow. That's the bench.
An opinionated 5-step assessment to figure out whether you're ready to embed AI engineers — and what you should do first. Written from ten years of placing engineers.
A properly scoped AI use case has a measurable outcome, a known data source, and a human you can name who will use the output. "We should do something with AI" is not yet a use case. This difference matters before you hire.
We will email it to you. No nurture sequence, no sales call — unless you ask for one.