OMKAR MAHALE
"37 trillion cells. Zero lines of code."
Who am I
What I've Shipped
Beyond the Resume
Find Me
Education
BE in Mechatronics Engineering & Industrial Economics
Mumbai University
Experience
- Identified hiring market gap → built & shipped Skopus AI (3-feature live ecosystem) from concept to active user base in under 12 months.
- Led end-to-end product: user research → PRDs → UI/UX design → full-stack development → shipped features with design consistency across the ecosystem.
- Running a 6-person cross-functional team while personally owning engineering, GTM strategy, financial ops, and stakeholder communication.
Education
MS in Financial Technology and Analytics
Stevens Institute of Technology
Experience
Quantitative Research Analyst: Detecting Rare Liquidity Events
Stevens Institute of Technology
Hoboken, NJ
June 2025 – Present
- Conducted equity research across S&P 500 equities, translating complex statistical findings into clear, actionable investment insights for research and trading teams across a $50M+ portfolio.
- Reduced signal noise by ~18% through rigorous analysis and model validation, improving the quality of decisions made by senior stakeholders.
Experience
Quantitative Research Analyst: Commodities & Equities
VTask Investments
India
June 2023 – Dec 2024
- Developed a Human-AI hybrid asset allocation framework that blended quantitative signals with analyst judgment, contributing to a ~40% improvement in portfolio performance.
- Analyzed commodity and equity market data to identify sector rotation trends and execution patterns, delivering findings that directly shaped trading strategy decisions.
- Managed client relationships by translating complex market analysis into clear, confident narratives, building trust with stakeholders across varying levels of financial sophistication.
GovBoard AI
- GovBoard AI simulates an enterprise boardroom: input any AI deployment decision, and four autonomous agents — CTO, CFO, Legal Counsel, Compliance Officer — each reason through it independently from their role's constraints and risk lens.
- Agents debate in real time, surface conflicting priorities, and converge on a synthesized governance recommendation with full reasoning chain — stress-tested before anything hits production.
- Built on LangChain agent orchestration, GPT-4 function calling, RAG-based role context injection, REST API backend, and a React/Zustand frontend with real-time streaming output.
ARIA — AI Financial Adviser
- ARIA is a personalized financial advisory agent — connect your portfolio report, and ARIA runs a full diagnostic, ingests your holdings into a live dashboard, and delivers real-time optimization recommendations stress-tested against current market conditions and macro signals.
- PDF/CSV portfolio ingested via LangChain document loaders → holdings vector-embedded in FAISS for context retrieval → GPT-4 function calling triggers fundamental analysis → live market data pulled via Yahoo Finance / Alpha Vantage APIs → position-level advisory streamed to frontend via WebSockets in real time.
- Stack: LangChain · GPT-4 · FAISS · Yahoo Finance API · Alpha Vantage · FastAPI · WebSockets · React · Zustand · Python · Cursor · Claude Code
B.O.D AI
- B.O.D AI is a boardroom simulation engine — input any business problem, and four iconic executive personas (Elon Musk, Tim Cook, Julia White, Ashley Bacon) debate it in real time, each reasoning, arguing, and deciding in character — output is boardroom-grade strategic decisions with full reasoning chains.
- Each persona runs as an independent LLM agent with role-specific system prompts and memory context; debate rounds are dynamically controlled by decision complexity — agents pass arguments sequentially, challenge prior reasoning, and converge on a synthesized outcome after N rounds; persona memory and context stored and retrieved via FAISS vector store.
- Stack: LangChain · GPT-4 · FAISS · Pinecone · Multi-Agent Orchestration · Dynamic Prompt Chaining · Python · FastAPI · React · Zustand · Cursor · Claude Code
Skopus AI
- Skopus AI replaces passive job-board browsing with real-time market demand signals — three-layer platform: Skopus (demand tracking), Skopus Verify (skill verification), and Skopus Hire (B2B recruiter discovery).
- Market signals ingested via AI pipeline → profiles vector-embedded in FAISS for semantic matching → GPT-4 career copilot delivers demand-aware guidance → recruiter discovery layer surfaces verified candidates to B2B accounts in real time.
- Stack: Next.js · React · Zustand · GPT-4 · LangChain · FAISS · FastAPI · Supabase · Tailwind CSS · Cursor · Claude Code · OpenAI Codex
- GTM: Owned full GTM motion — ICP definition, outbound sequencing, partner outreach, 500+ waitlist, $1.5M offer declined.
FlashBet
- FlashBet is a live prediction market protocol on Monad blockchain — real-time sports moments (penalties, corners, free kicks) trigger instant YES/NO micro-betting markets on-chain, resolve in 60–90 seconds, and pay out winners directly to wallet. No sportsbook, no house edge, pure Web3.
- Scripted match feed detects bet-worthy events → oracle wallet creates parimutuel YES/NO market on-chain → 60–90s betting window opens → users place bets with MON → window closes, oracle resolves outcome → smart contract calculates payouts → winners claim instantly; every step — creation, bet, resolution, claim — is a live on-chain transaction on Monad Explorer.
- Stack: Solidity · Hardhat · Monad Testnet · BetMarket (parimutuel engine) · MatchRegistry · OracleManager · Node.js · ethers.js v6 · WebSockets · Next.js 14 · Tailwind CSS · wagmi · viem · Zustand
Écho
- An AI recruiting agent that generates its own persona from a company's context, then engages candidates via a classify→decide→write reasoning pipeline using Claude's tool-use API — reasoning stays structured, not buried in text.
- Supports real-time voice interaction with a live camera preview for an interview feel.
- Stack: Next.js · TypeScript · Anthropic API · Claude tool-use · real-time voice · live camera. Deployed to production in 24 hours as a technical assessment.
A few things the résumé leaves out
My toolkit: Cursor, Claude Code, OpenAI Codex, LangChain, Supabase, Figma, Framer, Notion, Clay — dangerously productive when they're all open at once.
Fluent in Python, TypeScript, React, SQL, and the universal language of getting things shipped.
From market signal to PRD to deployed product — quant instincts, product thinking, and full-stack execution. One person. No handoffs.
Used 127 million Cursor tokens — in 2 weeks — building Skopus AI. Every single one was intentional.
Bought my first stock at 15. Been reading markets ever since.
Four years of live options, futures, and fundamental market research — not a hobby, a discipline.
Built without a check — earned cloud credits from AWS, Cursor, Google, and Supabase through pure GTM execution.
Coffee makes me superhuman.
Cooking makes my soul happy.
Omkar Mahale
Founder & Tech Lead · Quant × Product
Let's build something worth shipping.
Founder, quant, and full-stack builder. I read every message — pick a door.