2005 – 2010 · Mumbai & Pune
The Foundation
It started at Oracle Financial Services Software, building interfaces for the FLEXCUBE retail banking core ,the unglamorous plumbing that let a customer move money over the internet, a mobile app, or a phone call to a bank's IVR. Fund transfers, standing instructions, term deposits, direct debits. Nobody outside the bank ever saw this code, but if it broke, everybody noticed.
From there to Tata Consultancy Services, where I shipped a project end to end for the first time ,resourcing, permissions, timesheets, approvals ,not as a tester bolted onto someone else's build, but as someone who understood the whole shape of the system. That distinction would matter more than I knew at the time: you can't tell whether something is broken until you understand what it was supposed to do.
2010 – 2015 · Cognizant & Persistent Systems
Mastering the Craft
This is where "quality" stopped being a phase at the end of a project and became a discipline of its own. Inside Cognizant's Automation Centre of Excellence, I learned a dozen different tools, Tosca, Selenium, UFT, SoapUI, CA LISA for service virtualization ,because the right tool depends entirely on what's actually at risk.
Two projects from that era still come to mind first. At Credit Suisse, automating a cash control model with fifteen interlocking components, each one converting cash flows into payments ,get any one of them wrong and money moves to the wrong place. At JPMC, the Mortgage Express System: an entire loan's life, from the moment someone applies to the moment the loan closes, traced end-to-end. Then at Persistent Systems, an FX trade platform ,defining the business rules myself, not just testing someone else's.
What I took from those years: in finance, "it works" and "it's correct" are different questions, and most outages live in the gap between them.
2015 – 2017 · Pune
The Detour
I co-founded Decor Dreams, an interior design execution and marketplace business ,about as far from automation frameworks as a career can swing. No QTP, no test plans, no CI pipeline. Just clients, vendors, budgets, timelines, and the very physical reality that a wall either gets painted on schedule or it doesn't.
It taught me things engineering organizations rarely teach directly: how to build something from nothing, how to operate when there's no test environment to fall back on, how a business actually survives on trust and follow-through. I came back to technology a better systems thinker for having left it for a while.
2017 – 2019 · Danalitic
Back to the Code
A return to engineering as a Senior Software Engineer ,re-grounding in hands-on development before stepping into the role that would define the next phase: leading quality, not just practicing it.
2019 – 2023 · GoTo Group (Gojek), GoFin
Scaling Quality Across Continents
Four years leading quality engineering, test automation, performance, and customer experience across Gojek's financial products ,lending, payments, insurance, and collections. This is where the scope stopped being "test this feature" and became "make quality hold together across an entire financial platform, across three countries."
I architected automation frameworks for APIs and Selenium UI that pushed regression automation coverage past 80%, built CI/CD pipelines with WireMock and Grafana observability that cut deployment validation effort by 40%+, and built and led a distributed team of 30 quality engineers across India, Indonesia, and Vietnam ,setting the engineering standards and delivery governance the whole org ran on.
The Full Cycle Test that once took 54 hours per release came down to 9 ,an 83% cut.
Not by working faster ,by rethinking what actually needed to run, every single time.
2025 – Present · Incub8 Engineering
The Agentic Era
Somewhere in those years of frameworks and pipelines, a question kept surfacing: what if the system didn't wait for someone to write the test plan? What if, the moment a requirement landed, an AI agent started reading it ,acceptance criteria, API docs, past incidents, git history ,and began asking the questions a senior engineer would ask?
"These aren't test scripts. These are quality conversations."
That question became the work. As Principal SDET at Incub8, I architected and launched three flagship AI platforms: Omni Quality Platform (omni.vriddhix.ai) for autonomous AI test automation, StudioAgent (studioagent.vriddhix.ai) for autopilot short-form video generation, and SaarthiOne (saarthione.vriddhix.ai) for WhatsApp conversational commerce and bookings.
An AI-driven Agentic STLC ,multiple LLM-based agents, wired into DevOps and CI/CD, cutting testing effort by 30–40% and release validation time by 25%. Multi-agent frameworks with prompt engineering, deterministic guardrails, and validation orchestration, built specifically so the system never hallucinates its way past a real defect.
Alongside it, I've kept building in the open: Orbis Quant Agents, a LangGraph framework where AI analysts debate each other ,bull versus bear ,before a decision gets made, stress-tested for the same hallucination and bias risks I guard against at work. And a fully autonomous content pipeline where multiple agents research, write, evaluate, fact-check, and publish without a human touching the keyboard.
"We're moving from building automation frameworks to building intelligent quality systems."
The thread connecting all of it is the same one from FLEXCUBE in 2005: understand what the system was supposed to do, before you decide whether it's broken. The tools changed completely. The question never did.
Still going
What I've Learned Along the Way
Every experience has the potential to enhance you, if you handle it with a certain sense of awareness.
Twenty years in, that's still the closest thing I have to a guiding principle ,in code, in leadership, and in the long, unscripted parts of a career that never go quite the way the plan said they would.