experience

I worked on the partnerships team at DevRev and learned a lot from the experienced PBMs around me. Sitting next to people who’d been in B2B GTM for years made a lot of things click.

The biggest thing I took away was how you actually sell. In B2B, pitching a dreamy product isn’t enough. You sell a specific outcome — what changes for the customer, what gets unblocked, what number moves. Once I started thinking that way, the work got clearer: deal risk, mapping bugs, a renewal sitting in the wrong region.

A lot of my day-to-day was pipeline hygiene across AMER, APJ, and EMEA. I automated that with SQL analytics and internal AI agent workflows so at-risk opportunities didn’t stay buried. Those tools surfaced 13+ at-risk opps, including a $900K+ renewal. I also fixed 56 misrouted opportunities worth $1.2M+ in total pipeline.

On the partner side, I shipped a TypeScript partner-portal snap-in that repaired LATAM region mapping and region-based opportunity-owner defaulting. That recovered about 12 blocked deals that had been stuck on bad routing. I also helped run $4K in LinkedIn ad campaigns.

What stuck with me from DevRev is how much GTM depends on clean data, clear ownership, and a concrete customer outcome. That’s what makes the pipeline something you can actually operate.

At HaiIntel I was a forward deployed engineering intern, building AI tools for a hospital client across 5 workflows. Being forward deployed meant I was building for clinicians and staff with real patients and real schedules — people who didn’t have time for tools that only looked smart in a demo.

I indexed 8K+ patient records and cut retrieval time from about 10 minutes to under a minute. That one change taught me a lot about engineering: latency matters when someone is waiting on a record mid-shift. I also automated 300+ appointment follow-ups a month, which forced me to think about reliability, edge cases, and what happens when automation fails quietly.

What I learned most is that field engineering has to fit into an existing system without making life harder. You measure what the workflow actually costs people in time, ship the glue that makes the model useful, and keep iterating until the hospital team trusts it enough to use it without you in the room.