MS CS @ NYU + Stern FinTech β’ Graduating May 2026
1.5+ years across 4+ companies in AI, ML, quant finance, and full-stack, building systems that served 2M+ users, processed $10M+ in financial datasets, and sustained 95% uptime under 10K+ concurrent users. One of the earliest engineering hires at BetterCampus (0β1, then 1β10x).
I build the layer that keeps agentic pipelines alive when the environment breaks, detection, recovery, and self-repair baked in from the start, not bolted on after the first production incident.
Served as Advisor and Lead for Microsoft Learn Student Ambassador, Google Developer Groups, and Perplexity Campus Partner. Mentored 750+ students and volunteered with Smile Foundation. Competed and placed in 3x hackathons across 2,500+ participants.
AutoRFP β Menu URL in, signed deal out: 5 specialized AI agents negotiate vendor contracts autonomously using live CME/BLS pricing and OLS forecasting, no paid APIs required. β github.com/Utsavd7/AutoRFP-Automated-Ingredient-Procurement-System
QuantIQ India β Live NSE equity platform with NIFTY/BANK NIFTY indices, 6-tab analysis (technicals, financials, backtest, events), and a Groq-powered streaming research agent built for Indian market context. β github.com/Utsavd7/Event-Driven-Trading-Strategy-Simulator
NutriCam β iOS nutrition tracker with Gemini vision + custom YOLOv8, 90%+ accuracy in under 2 seconds, live on App Store. β github.com/turingsgarden/food-app β’ Download on App Store
Monte Carlo Portfolio Simulator β 10,000+ simulation paths with Sharpe/VaR/CVaR and efficient frontier visualization, 40% faster risk insights. β github.com/Utsavd7/Monte-Carlo-Portfolio-Simulator
Quantitative Risk Metrics Dashboard β Real-time VaR, stress testing, and correlation analytics with WebSocket live feeds. β github.com/Utsavd7/Quantitative-Risk-Metrics-Dashboard
pix2pix GANs / Microsoft Research β CLIP-guided diffusion with 96% precision and 30% diversity increase over 80K training steps under a Senior Director. β github.com/Utsavd7/pix2pix-GANs
I gravitate toward founding engineer and early-stage roles where the architecture decisions actually matter β if your agentic pipeline breaks silently at step 4 and nobody knows until a customer complains, that's the problem I'm wired to solve.
π§ utsavd7@gmail.com β’ π linkedin.com/in/utsavd7 β’ π utsavdoshi.vercel.app
From pix2pix GANs to 5-agent procurement pipelines: if it needs to stay alive under pressure, I've probably built the layer that keeps it running.



