Pricing · Rates · Risk · Systematic Research · Decision Systems
I build quantitative systems that turn market uncertainty into something you can price, hedge, validate, and explain.
M.S. Financial Engineering Candidate @ Stevens Institute of Technology
B.S. Computer Science, Magna Cum Laude @ Arizona State University
Jersey City, NJ · New York City Metro
Markets · Risk · Research Engineering · Financial Data Systems
I work at the intersection of quantitative finance, risk, and software engineering—building production-style research and decision systems rather than one-off notebooks. My work spans derivatives pricing, Treasury curve risk, portfolio construction, systematic signal research, credit and fraud decisioning, and event-market microstructure.
Each system runs from raw data through a reproducible Python pipeline to tested research outputs and a deployed interface. I am equally comfortable engineering the model, validating the result, and building the software that makes it understandable.
Does a signal survive reality? Five cross-asset strategies tested against costs, out-of-sample decay, parameter instability, regime shifts, and market stress.
walk-forward validation · cost-aware backtesting · no look-ahead · regime stress
Methods — Expanding-window research over 15 liquid ETFs from 2006 onward; signal-to-position lags; monthly rebalancing; cash fallback; 1/5/10/25 bps cost cases; train/test, walk-forward, parameter, and rebalance tests; stress windows spanning the GFC, Euro crisis, COVID shock, and 2022 rates selloff. Outcome: 3 signals survived, 1 was conditional, and 1 was rejected.
Proves — Research discipline and the willingness to reject a strategy that fails costs or validation—not just report the backtests that look good.
Live research report ↗ · Repository
A USD/MXN options book analyzed end to end—pricing, Greeks, hedging, tail risk, and P&L attribution.
Garman–Kohlhagen · Greeks · VaR / ES · Monte Carlo · QuantLib validation
Methods — Garman–Kohlhagen pricing; delta, gamma, vega, and theta; static spot hedging with residual gamma/vega exposure; 1-day 95% and 99% VaR/Expected Shortfall through 10,000-path joint spot-volatility Monte Carlo; full stress repricing; P&L decomposition into Greek channels plus residual; independent QuantLib benchmarking.
Proves — Derivatives pricing and the ability to take a hedged book apart into the risks that actually drive its P&L.
How an inflation regime breaks the textbook long-duration Treasury hedge—repriced cash flow by cash flow under a shocked curve.
DV01 / key-rate DV01 · convexity · real-rate / breakeven attribution · Vasicek / CIR
Methods — Ten curve-shock scenarios, including a measured 2022 replay; modified duration, total DV01, and key-rate DV01; exact versus duration-only and convexity-adjusted repricing; P&L attribution across real-rate, breakeven, duration, convexity, curve-shape, and residual channels; rule-based hedge overlays; 10,000-path Vasicek/CIR simulation with 1-year VaR/ES. The research pipeline closes with 40/40 validation checks passing.
Proves — Rates and curve analytics, plus the macro-risk decomposition needed to explain why a supposedly safe hedge fails.
Live rates engine ↗ · Repository
Four allocation rules, three crisis windows, and eighteen years of regime change across a 13-ETF multi-asset universe.
Equal Weight / GMV / Max Sharpe · regime-aware · drawdowns · factor exposure
Methods — Equal Weight, Global Minimum Variance, Max Sharpe, and rules-based Regime-Aware allocation under common constraints; maximum-drawdown paths; 95% VaR/CVaR; turnover and cumulative cost drag; normal-versus-defensive correlation matrices; effective independent bets; Fama–French factor-exposure diagnostics.
Proves — Portfolio construction with a clear view of correlation instability, crisis behavior, and the gap between in-sample optimality and implementable allocation.
Turns borrower, account, and transaction evidence into underwriting decisions—not merely model scores.
expected loss · policy thresholds · fraud controls · model-risk validation
Methods — Risk scoring and expected loss (EL = PD × LGD × EAD); approve / manual-review / decline thresholds; policy trade-off simulation; fraud and payments controls; portfolio monitoring; model-risk validation; transparent labeling of real and synthetic inputs across credit, card-transaction, stablecoin-flow, and macro data.
Proves — Credit and fraud decisioning as a complete system: evidence → model → policy → defensible action.
Live decision engine ↗ · Repository
Treats the official 2025 Formula 1 season as a portfolio of tradable binary event markets.
event markets · binary contracts · mark-to-market · microstructure · settlement
Methods — Real-season replay across 24 races and 500+ contracts; telemetry-driven probability repricing; circuit breakers; race settlement and championship futures; a $100,000 sandbox book with mark-to-market P&L, exposure limits, drawdown monitoring, Sharpe analytics, and a complete transaction audit.
Proves — How information becomes price: event-driven trading mechanics, market microstructure, and live portfolio risk inside a self-contained market environment.
Live market simulator ↗ · Repository
Research engines, not toy notebooks. Decision systems, not static dashboards.
The hard part is not computing a Sharpe ratio, a VaR number, or a credit score—it is proving the number deserves to be trusted.
- Validation first. Results face transaction costs, out-of-sample windows, stress scenarios, and independent checks before they are reported.
- Explainable by construction. Assumptions stay visible; P&L is attributed to its drivers; policy decisions can be traced back to evidence.
- Reproducible. Python pipelines generate deterministic research artifacts from documented data sources—no opaque browser-side calculations or black boxes.
- Engineered end to end. Research code, tests, data contracts, frontend, documentation, and deployment are treated as one system.
- Readable. Every engine ships with an interface that makes the analysis legible to someone who did not build it.
Research & Modeling — Python · NumPy · SciPy · pandas · scikit-learn · QuantLib
Quant & Risk — Garman–Kohlhagen · Greeks · Monte Carlo · VaR / ES · DV01 / KRD · Vasicek / CIR · portfolio optimization · walk-forward backtesting · stress testing · expected loss
Data — Yahoo Finance · FRED · NBER · Cboe · Kenneth French Data Library · reproducible CSV/JSON pipelines
Interfaces — TypeScript · React · Next.js · Tailwind CSS · Recharts
Also — C++ · R · MATLAB · Bloomberg and LSEG market-data exposure
Workflow — Git · GitHub · testing · Netlify · continuous deployment
- FRM Part I Candidate — August 2026
- Open to Fall 2026 quant co-op / internship opportunities
- Targeting full-time 2027 roles across quantitative research, trading, derivatives, risk, and quantitative engineering
- Current research interests: systematic-strategy validation; derivatives, rates, and portfolio risk; credit, fraud, and payments decisioning