Skip to main content

Madison · CT

UW–Madison ’27

Data Science & Economics

Technical product builder. I take ambiguous financial, billing, and AI workflows and ship them as products with measurable outcomes.

Now

  • Building: TAM, from discovery calls with diligence practitioners
  • Just shipped: Ghostkeys at Build Fest
  • Studying: Final year of Data Science & Economics at UW–Madison
  • Running: An automated trading system on my own portfolio

(01)Work

Selected work

05 projects, one frame each
2025–2026

Each opens in place as a case study

(02) Principles

How I build

Five rules I keep coming back to, each learned on a real project. Open one to see the work behind it.

01 / 05

Finance has to be able to trace every dollar. I keep the core logic plain and checkable, and use models only where judgment actually helps.

Evidence

Zendesk DIB: matching core in Snowflake SQL; AI for classifying mismatches and simulating users.

02 / 05

Ramp and ELA contracts kept breaking the logic. Instead of stalling everything, I shipped every other deal type and phased those into a follow-on release.

Evidence

Zendesk DIB: in production, saving 9,000+ analyst hours and $2–⁠3M a year.

03 / 05

Before building more of TAM, we sat with due-diligence practitioners and Managing Directors and turned what they said into the roadmap.

Evidence

TAM: discovery with practitioners and MDs at EY and PwC.

04 / 05

AuraHealth phones patients after a visit and sorts what it hears by urgency. What happens next is the doctor’s call: anything urgent reaches them by text.

Evidence

AuraHealth: high-urgency follow-ups text the doctor and flag the case. Google Award, CheeseHacks 2026.

05 / 05

A prototype teaches you what you already believed. An App Store listing with reviews teaches you what people actually do.

Evidence

Flux: live on the App Store, 350 users, 4.8★ rating.

(03) Log

Changelog.

Release notes, newest first. v2023.07 → v2026.09

Showing all 17 entries

  1. NowRUNNING

    An AI trading system

    Agent-driven strategy execution, rebalancing, and risk controls on my own portfolio.

  2. 2026.09BUILT

    Ghostkeys

    At Build Fest, with a team.

    OpenCode
  3. 2026.08SHIPPED

    Data Integrity Bot, to production

    Quote-to-Cash reconciliation at Zendesk.

    Open
  4. 2026SHIPPED

    Two more internal AI tools

    A token optimizer that cut AI spend by ~20%, and a documentation-freshness checker.

  5. 2026.06JOINED

    Zendesk

    AI Product Intern, Business Operations.

  6. 2026ACCEPTED

    Summer AI Lab (SAIL), UW–Madison

    An agentic AI program co-sponsored by OpenAI.

  7. 2026WON

    Google Award, CheeseHacks

    For AuraHealth.

    OpenCode
  8. 2026WON

    2nd Place, MadData

    For TAM.

    OpenCode
  9. 2026BUILT

    HelloNeighbour

    Anonymous voice notes between UW–Madison dorm neighbors, matched by a KNN written from scratch. ClaudeHacks.

    Code
  10. 2025.06JOINED

    EY

    M&A Consulting Intern. Financial due diligence on a live $78M M&A deal.

  11. 2025WON

    1st Place, Cursor Hackathon

    For EarningLens, which fact-checks earnings-call guidance against SEC filings within 3 seconds.

    Code
  12. 2025STARTED

    TAM

    Co-founder & CEO.

    Open
  13. 2025STARTED

    Flux

    Co-founder. Live on the App Store.

    Open
  14. 2024.09STARTED

    UW–Madison

    B.S. Data Science & Economics.

  15. 2023.07JOINED

    Synechron

    AI Engineering Intern. A claims-cost model that cut manual processing 70%.

  16. —SHIPPED

    BoxMate

    A student storage marketplace. $10K in revenue in its first two weeks.

  17. —BUILT

    A stock trading system

    A PAMR mean-reversion engine with LLM market commentary.

(04) About

Yash, briefly.

I turn ambiguous financial, billing, and AI workflows into shipped products, then measure what they changed.

Based in Madison, Wisconsin. Before co-founding TAM, an AI due-diligence platform, I did financial due diligence on a live M&A deal at EY.

Education

UW–Madison, B.S. Data Science & Economics, May 2027

Deep Learning & Generative Models · AI/ML · Database Systems (SQL) · Algorithms · Linear Algebra · Statistical Computing (R)

Toolbox

Product
Product discovery · Requirements definition · Prioritization & scope trade-offs · Roadmapping · Cross-functional collaboration · Stakeholder management up to VP/SVP · GTM strategy
AI & ML
Claude API · LangChain / LangGraph · RAG · Agent orchestration · Tool use · Prompt & eval design · NLP / NER · FinBERT · spaCy · PyTorch · scikit-learn
Engineering & Data
Python · SQL · R · FastAPI · REST APIs · PostgreSQL · ChromaDB · Snowflake · Redis / Celery · Docker · Git
Systems & Analytics
Jira · Confluence · Salesforce CRM / CPQ · Zuora · Tableau · Power BI · Advanced Excel · Financial modeling · DCF valuation

Lately

84 contributions in the last yearLast push: Personal-Website · GitHub

Off the clock

  • A STEM lab in rural India

    Founded and equipped it for an underserved rural community: computers, robotics kits, and a hands-on curriculum, run with local teachers as an ongoing program. 100+ students so far, and still running.

  • Karate

    Black belt, earned over years of training.

  • Student founders

    I mentor early-stage founders in UW–Madison’s entrepreneurship community.