Kavin Uppal

Atlanta, GA

My passion lies in building. I like taking something that starts as a thought and making it real, and I like what comes after even more: watching people use it, then reshaping it until it actually fits them.

I don't think building the product and understanding the user are two separate jobs. Most of what I've worked on has been an attempt to hold both at once. Chasing the overlap between what gets built and what someone actually needs is probably impossible, but I think that chase is what makes a product worth building.

01Now

Finishing dual master's degrees at Georgia Tech in Computational Science and Engineering and in Quantitative and Computational Finance, after an undergrad in Industrial Engineering.

02Experience

Capital One

Jun – Aug 2026

Product Management & Analytics Intern

Reframed how a major card portfolio measured cannibalization, surfacing value that the existing models had been counting as loss.

The Coca-Cola Company

May – Aug 2025

Product Strategy Intern

Built the framework that decided where recycling investment went, and aligned internal teams and independent bottlers behind a single strategy.

Millennium Management

Jan – Apr 2025

Project Based Software Engineer

Built a low-latency order management and matching engine, then demoed it to trading leadership and rebuilt it around how they actually worked.

McKinsey & Company

Sep 2024 – May 2025

Project Operator

Designed and ran the workshops that drove client adoption of new digital tools, working alongside C-suite leadership to diagnose where operations broke down.

03Selected work

Signum

2025–2026

A B2B pricing intelligence startup that simplified procurement by unifying pricing scattered across vendors into a single comparable view.

Python · PostgreSQL · FastAPI

Gradient-boosted and neural classifiers over a severely imbalanced transaction set, reaching 78% precision on held-out data.

Python · XGBoost · scikit-learn

Interactive analytics over Yelp restaurant data, built to make a large review dataset explorable rather than just queryable.

Python · D3.js

Georgia Tech Financial Services Innovation Lab

2024–2025

Automated a DCF valuation pipeline over weekly market data, and built the scraper that pulled and normalized financials straight out of SEC filings and earnings transcripts.

Python · Tableau · Excel

Paired electric vehicle adoption against air quality measurements to test whether the relationship holds where adoption is highest.

Python · pandas

04Principles I stand by

The grass is greener on both sides

Every decision costs you the one you didn't make, and the road not taken always looks better from here. It would have looked exactly the same way from the other side.

Everything averages out

Runs in either direction feel more meaningful than they are. The goal is raising the average, not chasing the highs or flinching at the lows.

Probability should dictate decisions

Losing hurts more than winning feels good, so failure gets weighted more heavily than it has earned. The probabilities are the only real information available, which makes them the only honest thing to decide on.

Luck is where hard work meets opportunity

Luck is where effort meets opportunity, and opportunity is not yours to control. When it is scarce, the only variable left is how much work you put in.

Success is not zero sum

It does not have to be taken from someone else to count. There is a version of the world where the people around you do well too, and wanting that costs nothing.

05Beyond work

Travel · Bollywood dance · iPhone photography · Running · Pickleball · R&B · Experimental cooking

06Elsewhere