Treasurer
F1@Purdue
The club's books get the same treatment as a pit-stop plan: every line accounted for, no time lost. Budgets, expense tracking, and fundraising with the exec board.
…and it's lights out and away we go.
I write code that understands markets.
Senior at Purdue, double-majoring in Computer Science and Economics. I've forecast the U.S. alcohol market for Molson Coors, classified 30 years of the economy into regimes, and modeled F1 pit strategy — because a pit-stop call and a portfolio call are the same decision. This page is one lap through how I got here.
Turn 01 The grid — where I started
I came to Purdue from Ahmedabad in 2023 as a pure CS major with exactly one conviction: I was not going to spend my life just coding 9-to-5. First semester, I took macroeconomics on the side — and watched the Fed move entire markets with a rate decision. How much people buy, borrow, and build, steered from one room. I was hooked. Econ became a minor, then a double major, and every class since has deepened the awe.
Since then, every project has been some version of the same move — take a noisy system (the U.S. economy, the alcohol market, a Grand Prix), put real data on it, and find the decision hiding inside. My macro dashboard classifies 30 years of the economy into color-coded regimes. It felt only fair to run the method on myself. Tap a chapter below.
* projected — out-of-sample, as every good forecast should be.
Turn 02 The race — experience
F1@Purdue
The club's books get the same treatment as a pit-stop plan: every line accounted for, no time lost. Budgets, expense tracking, and fundraising with the exec board.
The Data Mine × Molson Coors
Five-year forecast of the U.S. alcohol market. I owned Beer and RTDs — LASSO with macro drivers after SARIMAX and XGBoost flamed out on 35 annual data points. Headline finding: RTDs are recession-resistant.
Purdue University
Weekly office hours, grading, and supplemental materials — turning aggregate demand and monetary policy into things students actually remember.
Cygnet.One
Migrated legacy ASPX modules to ASP.NET MVC, built delivery-status and KPI dashboards, and integrated C# controllers with SQL stored procedures.
The Data Mine × AgReliant Genetics
Engineered 80+ features down to ~20, raced K-means against DBSCAN against hierarchical clustering, and delivered the story in Tableau.
Shyam Sir Classes — Ahmedabad, India
The summer before Purdue: taught Java to a class of 30 back home with a 100% pass rate — where I learned that explaining something clearly is harder than building it.
Turn 03 The garage — projects
Each one starts with a question worth answering.
What state is the economy actually in, right now?
Gaussian Mixture Models on 30 years of FRED data classify every month since 1996 into Expansion, Slowdown, Stagflation, or Recession — then map which S&P sectors win in each. Correctly flags 2008, 2020, and the 2022 stagflation.
What's the worst that can plausibly happen to this portfolio?
10,000+ simulated paths with Geometric Brownian Motion on real yfinance data — VaR, drawdowns, and probability distributions for SPY, QQQ, DIA, and single names.
When does pitting early actually win the position?
Real telemetry via FastF1, nine publication-quality visualizations — stint degradation, pit timelines, and a success-probability heatmap that reads like a strategy playbook.
Can you rewind a Grand Prix and study every decision?
Reconstructs entire races from telemetry into an interactive 25 fps replay — live leaderboards, per-driver speed/gear/DRS, qualifying deep-dives. Multiprocessing cut compute time 80%.
What's actually happening when you hit Enter in a terminal?
A full shell in C/C++: pipelines, I/O redirection, signal handling, wildcards, subshells, and an interactive line editor with history and tab completion.
Which customers behave alike, and what do they want?
Unsupervised segmentation on real agricultural sales data: K-means with silhouette-score validation, PCA for structure, Tableau for the business-facing story.
How do you make performance reviews not terrible?
MVC platform with role-specific interfaces for HR, managers, and employees — goal setting, KPI tracking, and reporting end to end.
Turn 04 Pit wall — the setup sheet
Turn 05 The podium — results
Selected to Purdue's team for the national monetary-policy competition. Built economic forecasts and policy recommendations, with a research focus on how policy hits different socioeconomic groups, housing markets, and labor force participation.
Reached the final round among 20+ teams university-wide in Purdue's startup pitch competition.
15-day global trading simulation with 12,000+ participants. Wrote Python trading algorithms across multiple asset classes, adapted to shifting market conditions, and outperformed the majority of teams on portfolio growth and risk management. The competition that pointed me at markets for good.
One of 12–15 students selected for advanced economic research at Purdue. Studied digital currency models and payment systems; presented findings to faculty and industry professionals.
Turn 06 Cool-down lap — off the clock
Turn 07 Final corner
That's the radio call to come in — consider it sent. Open to internships and full-time roles in fintech, tech, healthcare, and consulting. Or just a good conversation about regime models and race strategy.