Led the development of direct indexing and tax-aware portfolio optimization solutions with a range of client customizations and constraints. Developed standard and mixed-integer algorithms for portfolio optimization. Created sophisticated constraint attribution and holdings decomposition reporting, used within our InvestOS-compliant, AI-agent-friendly technology stack.

As a founding team member, built an institutional-quality asset manager from scratch — overseeing trading, risk management, and compliance for market-neutral and long-biased U.S. equity portfolios with over $100mm AUM, executing more than $15B in equity trades via quantitative, machine learning-driven strategies. Led a cross-disciplinary team of portfolio managers, data scientists, and technologists while managing prime broker relationships, chairing risk and best execution committees, and developing performance attribution tools and analytical dashboards.

Managed end-to-end portfolio optimization and trade list construction across equity, commodity, and currency portfolios in the U.S. and Asia, with responsibility for trade approvals and oversight of the alpha generation team's day-to-day operations. Developed a factor-based performance attribution model and a portfolio simulation environment used to calibrate models, investigate trading strategies, and support the development and marketing of new investment products.

Managed a team of eight quantitative software engineers in San Francisco and London, holding operational responsibility for the Global Markets Strategy Group's analytical and alpha generation systems. Started the integration of BGI's optimization technology into BlackRock's Aladdin platform, coordinating engineers and business analysts across New York and San Francisco.

As a senior member of the Proprietary Analytics Group, developed and maintained portfolio optimization code used to manage over $500 billion in assets, and built a widely-used characteristic portfolio construction and analysis engine. Conducted research on signal performance, liquidity, and portfolio constraints, and collaborated with Stanford researchers to develop efficient numerical techniques for multi-period mean-variance optimization.

Developed proprietary algorithms and numerical models for the strategic pricing and procurement of electronic components, incorporating consumer demand forecasting and product lifecycle and cannibalization analysis. Led the Workbench project — data visualization and model parameterization tools — from requirements gathering through implementation in support of client services.

Developed algorithms for forecasting, sampling, and statistical analysis of pharmaceutical SKU demand, and built a simulation environment to evaluate the efficacy of inventory management strategies. Originated a Latin hypercube sampling technique for demand analysis, presented at INFORMS 2000.

Wrote commercial software for semiconductor processing and device simulation, developing hybrid mesh generation techniques and level set algorithms for multi-dimensional finite element simulation and material interface tracking.

As a Ph.D. student, developed an original algorithm for simulating plasma-based semiconductor processing using advanced numerical techniques including Godunov methods for conservation laws, multi-grid solvers for elliptic equations, and finite volume schemes on locally refined grids.

Performed numerical simulations in support of pulsed power equipment design. Used lumped element and transmission line circuit models, developed finite element models for electrostatic and magnetostatic simulations, and generally assisted senior physicists with computational support.