Umar Darsot — Software Engineer

I'm a software engineer and a math co-op student at Waterloo. Most recently I was at Whop on wallets and growth engineering, shipping a virtual card program for creators and a migration funnel that reads a merchant's live fees to show them what switching would save. Before that I led the mobile experience at Dapital: a 0 to 1 iOS app in Swift and UIKit plus the real-time data layer underneath it — live prices for 150+ pairs, and an interface that stays correct when things move fast. Alongside Dapital, data and product work at Interac on the e-Transfer team. Before all of it, document automation at Purolator. Outside of work I'm usually shipping something small and sharp — a flight delay model that took first at I4, an MCP server that lets an agent drive an iOS Simulator, a parametric Blender script that prints a wallet sized to my own phone.

Experience

• Whop — Software Engineer, May 2026 — Aug 2026, Palo Alto, CA. Shipped Whop Cards, a virtual card program for creators — issuance, spend limits and freeze/cancel flows in Rails and Postgres. Scaled to tens of thousands of cards across thousands of creators. Built the Stripe migration funnel: a page that reads a merchant's live Stripe fees to show what they'd save by switching, paired with automated card import so subscriptions kept billing. Drove seven figures in migrated payment volume. Architected a Hyperliquid perpetual futures integration as a native Rails API resource, using non-custodial Privy embedded wallets, USDT collateral bridged via Plasma, and builder-code fee attribution. Worked on Apple Pay and Google Pay express checkout, extended the checkout API with dynamic required/recommended action resolution, and instrumented the card application funnel.
• Dapital — Software Engineer, Sep 2025 — May 2026, New York City. Built and launched a 0 to 1 iOS perpetual futures trading platform in Swift and UIKit, powered by Hyperliquid. Built the real-time data layer streaming live prices for 150+ crypto pairs over WebSockets and gRPC, refreshing charts, open positions and portfolio value across the app in under 80ms. Designed and built the perps candlestick chart, distilling the strongest interaction patterns from leading mobile trading apps into a single scrub-driven component that became the standard across the app. Owned the broader mobile trading experience: order entry, execution, portfolio updates and production reliability. Also built the web platform in Next.js.
• Interac — Product — Money Movement / e-Transfer, Jan 2026 — Apr 2026, Toronto, ON. Built a cross-bank intelligence system mapping which banks 9M+ Canadians actually use by linking e-Transfer and debit activity, packaged as an analytics product Interac could sell back to those banks. Ran network analysis across millions of banking transactions, surfacing 10K+ risky account relationships and activity patterns invisible inside any single institution's data. Built an AI-powered market intelligence pipeline processing 40+ fintech sources biweekly through a multi-stage LLM workflow — adopted across the e-Transfer and Debit teams, saving 6+ hours weekly. Scraped Reddit, X, RedFlagDeals and news, deployed on Railway, delivered via Telegram and HTML email.
• Purolator — Software Engineer, Data Science, May 2025 — Aug 2025, Toronto, ON. Built a document automation platform in Python, Flask and spaCy to extract and validate structured information from 50+ weekly operational reports. Modernized legacy internal systems into containerized cloud-ready services with Docker and CI/CD, improving deployment reliability through automated testing.

Education

• Bachelor of Mathematics — Financial Analysis and Risk Management, Co-op, University of Waterloo, Expected April 2029.

Projects

• Flight Delay Contagion (2026) — Ranks flights by how much lateness they propagate downstream, not by how late they are. Built with XGBoost, MILP, D3.js, Python.
• Project EVE (2025) — Multimodal conversational AI that lip-reads silent video and speaks back. Built with AV-HuBERT, Tacotron 2, HiFi-GAN, PyTorch, GPT-4.
• Claude App Flow (2026) — iOS simulator QA tooling as an MCP server, so an agent can drive and verify an app. Built with MCP, JavaScript, Facebook idb, iOS Simulator.
• Whop Tasks (2026) — AI-verified task marketplace where an LLM grades proof of work before money moves. Built with Next.js, TypeScript, Prisma, Postgres.
• UFC Fight Outcome Model (2026) — Scored against the closing betting line instead of against a coin flip. Built with scikit-learn, gradient boosting, Python, bootstrap CI.
• Job Application Pipeline (2026) — Automates ATS submissions end to end — resume generation, form filling, tracking. Built with Bun, TypeScript, Playwright, Postgres, Railway.
• Weather Edge (2026) — Fair-value pricing and fractional Kelly sizing for Polymarket temperature markets. Built with Python, NumPy, Polymarket API, Kelly criterion.
• Parametric Print Wallet (2026) — Blender/bpy script that generates a custom TPU phone and transit card wallet. Built with Blender, bpy, Python, 3D printing.
• ChopShop (2026) — Splits 3D models too big for the print bed into printable chunks with dovetail joints. Built with FastAPI, trimesh, React, Three.js.
• Market Radar (2026) — Scans Reddit, X, RedFlagDeals and news for Canadian fintech signals, then mails a digest. Built with Python, LLM pipeline, Telegram API, Railway.
• ESG Investment Screener (2025) — Screens 315 S&P 500 companies by ESG risk, sector and market cap. Built with Python, Streamlit, Pandas.
• Number Recognizer (2026) — A digit recognizer that measures its preprocessing gap rather than reporting an accuracy. Built with TensorFlow, Keras, Python.
• Churn Predictor (2025) — Keras MLP on Telco churn — 77.9% test accuracy, 47.3% recall on the churn class. Built with TensorFlow, Keras, scikit-learn.

Skills

• Languages: Python, TypeScript, JavaScript, Ruby, Swift, SQL.
• Backend & Data: Ruby on Rails, FastAPI, Flask, Node.js, Postgres, MySQL, BigQuery, gRPC, WebSockets, REST APIs.
• Frontend, Mobile & Infra: Swift, UIKit, SwiftUI, React, Next.js, Tailwind, Docker, CI/CD, Railway, Git, Linux.
• Fintech & ML: Stripe, Privy, Hyperliquid, USDT/Plasma, Ledger reconciliation, Card issuing, XGBoost, GPT-4, NumPy, Pandas, spaCy.

Contact

udarsot@gmail.com · GitHub · LinkedIn