Harsh.

A littleAbout me

Hi! I’m Harsh from Winnipeg, Canada. I started programming at 16 when I volunteered to help my high school teacher redesign our school website, and I’ve been hooked ever since (fun fact: we wrote a large part of it in Java applets… whew, thank god for JavaScript).

The web is what made me love software, and I care deeply about taste, performance, and the tiny details that usually get an “oh, who’s gonna notice that?”

Currently I’m at Fiscal AI, where I work across the product, API and build way too many financial charts.

… and a lot moreAbout me

Where I am

Winnipeg, Canada

--:--Winnipeg, me
--:--you
--hours apart

How I got here

The short story

I’ve always been a tinkerer, so it’s no surprise to the friends and family who knew me growing up when they learn what I do (although I’m not sure my parents actually know what I do). My tinkering with computers started, of all places, with the internet.

I was enamoured with online multiplayer games. It felt like magic: each of us at home, and somehow all together. I wanted to learn everything about how they did it. I didn’t know it at the time, but this curiosity was building an appreciation in me for engineering and the web as a whole.

This obsession with figuring things out stayed with me, and in high school it got me a chance to actually build on the web, not just learn about it.

After graduating I went deep on web development at Terracor, where over three years I learned to build great software, leading everything from e‑commerce builds and marketing campaigns to ERP integrations, before moving on to Fiscal AI.

What I do now

Fiscal AI

Fiscal AI is a financial data provider that powers some of the largest financial platforms and brokers around the world (if you’re in the financial corners of social media, you may have seen our charts floating around). I joined the team to work on the terminal product, our all‑in‑one research platform, which provides everything from financial analysis and an investor relations feed to stock screening. Over the past six months I’ve expanded my role to cover our data infrastructure, and I now lead the ownership and ETF & funds datasets.

I’ve touched many different areas of the product at Fiscal AI, but my favourite has to be the organization management platform I built alongside the launch of our data API. It quickly became one of the most critical pieces of software in the product: it’s not only the entry point for every customer onboarding onto our platform, but also the layer that handles entitlements for every API and MCP request a client makes.

Fun facts

  • I was recently part of an onboarding revamp that increased the user conversion rate by over 30%.
  • I was the first member of our engineering team to work across both our products at the same time.

What I believe

Some opinions

  • 01On buildingYour products should excite you; dogfooding is a strong sign of this.
  • 02On experimentingIt’s not enough to know what works; you also have to understand why it works the way it does.
  • 03On workingAim to make the team better, not to be the star on it. A better team outlasts a better you.

What I make

Own work

  • LynqPrivacy-first web analytics with AEO traffic attribution, UTM tracking, A/B testing, Core Web Vitals, errors, and uptime monitoring.
  • WritingUnnecessarily large deep dives into all things web.

Let’s takeA break

People are walking in to learn a skill that solves their problem. Can you get them into the right course?

Experts

And finallyWhat I’d build

Questions.

People rarely start with a course in mind. They start with a question: “how do I price a B2B product?”, “how do I get my team to use AI?”, “why does my onboarding lose people on day two?” And the answer is usually somewhere specific: a week 3 lesson, or twenty minutes into a free session where an expert took that exact question live.

Maven could let people search with the question itself. Every syllabus, lesson, and recording would be split into the moments where something gets answered, so a question lands on the lesson or the exact minute an expert answers it, playing from there, with the course it comes from right below.

And once questions are what people search with, they’re worth keeping: they show which courses answer what people are asking, and experts can see the questions that keep coming up and teach to them in their next cohort.

Catching up.

Falling behind in a live course is easy; it’s rarely by choice. There’s a full-time job, a deadline at work, something at home. Catching up becomes the hard part: hours of recordings, chat threads to follow, and a project to build on your own.

Maven could put together a catch-up for each person from what they actually missed. It knows which sessions you weren’t in and what’s due next. So it pulls out the parts of the recordings this week’s project depends on, what the cohort worked out in the chat, and the links that were shared, each one opening the recording at the moment it happened.

There’s a lot more I’d love to talk about, and I hope this starts that conversation.