cat ~/about.md

The short version

I spent five years running global marketing campaigns for Fortune 500 clients, then spent the last two building AI systems on a home server in my bedroom. Both are true at once, and the second grew directly out of the first.

The MarTech years

I started as a Campaign Analyst at Epsilon in 2021, straight out of a BSc, building HTML/CSS email templates and segmentation logic on Braze and EPCM. By 2023 I was a Campaign Business Analyst running lifecycle programs for Marriott, AbbVie, McKinsey, and National Geographic — personalization logic, QA gates, SLAs across time zones, the full apparatus of enterprise marketing ops. I picked up a Magnitude Award and two CMS Excellence Club wins along the way, but the real education was operational: what breaks when a client-facing send has a data error 24 hours before go-live, what “done” means when three teams and two vendors touch the same campaign, why QA discipline isn’t optional once you’re at scale.

The switch

I didn’t wake up wanting to write code. I noticed that the tools running these campaigns — segmentation engines, personalization logic, approval workflows — were themselves automation, and that the more interesting version of my job was building that layer, not operating inside it. So I started building. First inside Epsilon: a RAG chatbot (n8n, OpenAI embeddings, GPT-4o mini) so the ops team could stop searching SOPs by hand. Then real production tools outside work — an invoice-processing pipeline for a local pharmacy, a WhatsApp bot, a fitness coach, a CCTV alert system. No bootcamp, no course certificate. Three years in, the MarTech background isn’t baggage I’m shedding — it’s why I understand what a stakeholder actually needs, and why everything I ship gets a test suite and an audit log, not just a demo.

How I work

I run a home lab — Ubuntu server, Docker Compose, n8n, Cloudflare Tunnel, Tailscale — hosting things people actually depend on daily, not weekend projects that die after the blog post. The pharmacy invoice pipeline has 116 tests and has been in daily production use since launch. I use Claude Code and other AI coding assistants deliberately, as a force multiplier on architecture and debugging — my Python and JS are working-knowledge, not academic, and I don’t pretend otherwise. What I bring is judgment about what to build and whether it’s actually working, which is a different skill than syntax.

Now

I’m still at Epsilon by day, still shipping production automation by night, looking for where that combination — engineering plus operational scar tissue — is worth more than either alone. Based in Mangalore, India, IST, set up to work async with US and EU teams.