2014–2019 · YEXT · FIVE YEARS

Learning what a pipeline costs

Yext keeps business information correct everywhere it appears online. It was my first job out of Carnegie Mellon and I came in as a full stack engineer: Java and Python on the back end, JavaScript on the front, pipelines running on EC2. My team owned analytics.

I owned the pipeline. It pulled events from the Google and Facebook APIs and from our own pixels on partner sites, so when someone searched on Yelp the event landed with us. Then I rebuilt it, moving it off Python and into a service in our Java codebase that ran on its own. Every customer dashboard was reading from it, so none of that work could be done loudly.

I built the dashboard sitting on top of it too: live charts, scheduled email reports, an activity log. I led the local rank tracker, which followed a client’s Google ranking for a keyword over time. By the end I was tech lead of four engineers on external data sourcing, which was the first time I was responsible for anyone’s judgement but my own.

2019–2024 · WELLFOUND (FORMERLY ANGELLIST TALENT) · FIVE YEARS

Search, scale, and the parts nobody volunteers for

I left Yext because the engineering team had gotten big and I wanted my decisions to carry further. I wanted a different job description too. Product engineer was the appeal: get handed a problem and own it end to end, from validating the idea through to shipping the real thing. Wellfound is where startups and the people who want to work at them find each other. I was there five years, mostly in Rails and React, with a few Node and TypeScript services, Elasticsearch and Postgres underneath.

I started on company profiles. Most of them were filled in once and then left to go stale, which is no help to a candidate trying to work out whether they want to work somewhere. I built a service that took a company name, went out to the web to find it, checked it had the right company, and imported what it came back with. Profile completeness went from around 20% to 40%.

Then Solr. It was legacy in every sense: slow, undocumented, and taking the site down about once a week. I replaced it with Elasticsearch one endpoint at a time. Build the replacement, run it, watch it for a fortnight, move on. It took a few months and nobody ever saw it happen, which was the point. Search became something we could build on again.

After that I put my hand up for Trust and Safety and ran it alone for nearly two years. Strategy, roadmap, the KPIs I set and then had to hit, the engineering, and the relationships with our operations and support teams inside and recruiting partners outside. The core of it was a system for catching bad actors: fake candidates, spam applications, companies posting jobs that did not exist. It halved moderation time, taking a full day of someone’s work down to about four hours, with a person still reviewing at the end. I added candidate and company verification on top of that.

The last stretch was back on product, building easy-apply. We crawled Greenhouse, Workday and others, transformed the listings and brought them into our system. Then the more ambitious half: a year-long pilot where we submitted the application for the candidate. They answered the questions with us, and a system behind the scenes opened the real listing and filed it on their behalf. It handled more than two million applications at a 99% SLA. Alongside all of it I moved our Salesforce syncing over to Hightouch without dropping anything, and worked on the data team behind the company’s AI initiatives.

2024–2026 · HEADWAY · TWO YEARS

Leading, and building the thing at the same time

Five years in, the marketplace had gotten harder to work in and I’d grown about as much as I was going to. A recruiter got in touch about Headway. I talked to enough people there to believe in it and went. Headway does two things that turn out to be one thing: it makes it easy to find a therapist who takes your insurance, and it makes it possible for that therapist to get paid by the insurer.

The engineering team was about fifty when I joined, eight of them on the patient side, which had barely started. That was the appeal. I led and mentored six engineers in the Patient Growth pod, owning the architecture and the execution of how patients find their way in, across paid and organic channels.

Most of what I built was foundations: the internal tooling that makes a company good at SEO and search marketing rather than lucky at it. A sitemap generation service. Google Ads API integrations. A provider-level Google Ads platform from nothing, covering campaign and ad group generation, keyword strategy, budgets and performance tracking.

The one I’d point to is the page generation system, which builds our SEO and paid landing pages out of statistics, metadata and content pulled from our internal systems. The whole design problem was the gap between launching two pages and launching several hundred. Over my two years the patient side grew more than 200% year over year. It’s also the first thing I’ve built where the output is measured in people who found care.

NOW

I left Headway after two years for family reasons rather than professional ones, and moved back to India. I’ll be between there and the US, mostly there. It’s the first gap in twelve years I didn’t plan, and I’m treating it as one: working out what I want to do next rather than which offer to take.

Some of that is building things of my own, and finding out whether starting something is actually in me. The rest is staying around people who are good at this.

If you’re working on a problem you care about, big or small, and think I could be a fit, I’d like to hear about it. I’m open to full-time roles, and also to part-time, contract and fractional work, including as a thought partner.

I’m at hello@vishaljeet.com, or on LinkedIn. A short note is plenty.

V