LLM Trading Systems
Deployed decision pipelines across FX, commodities, and options, including DSPy/GEPA prompt optimization and multi-model reasoning over market narratives.
DSPyGEPAOANDALLM inference
About
From research idea to production system
I build and operate AI systems end-to-end—from multimodal retrieval and tool-using agents to live quantitative trading platforms. My work spans model adaptation, evaluation, APIs, cloud infrastructure, and the production reliability required to keep automated systems running.
1.5+
Sharpe, agent-mined strategies
5%
Weekly returns on capital at risk
Selected Systems
Production AI across models, markets, and infrastructure
ML Trading Systems
Built Bayesian volatility signals, an HMM/NSGA-II equity portfolio, and a Chronos-LoRA plus TabPFN interday options platform.
PyTorchChronosTabPFNNSGA-II
Agents & Multimodal Retrieval
Built an MCP signal-mining agent and a full-stack PDF research assistant with contrastive CLIP tuning, Qdrant retrieval, and vision-LLM synthesis.
MCPCLIPQdrantFastAPI
Production Infrastructure
Shipped containerized, scheduled systems with persistent state, cached model artifacts, broker integrations, failure recovery, and automated deployment.
AWS LambdaModalDockerGitHub Actions
Experience
Building systems that leave the notebook
Asset Management Firm · Stamford, CT
LLM-backed trading systems
- Deployed an intraday FX pipeline that synthesized RSS news and price action into trade decisions, executed through OANDA, and closed the research-to-production loop with an evaluation-gated DSPy/GEPA prompt optimizer that promoted a new prompt only when it outscored the incumbent on archived outcomes.
- Deployed a commodity workflow combining macro news, price action, and sovereign yield-curve differentials through a mixture-of-experts pipeline that assigned each model a role based on its observed reasoning bias, with conflict-resolution and abstention rules.
- Implemented an LLM-driven intraday options strategy that translated live stochastic-volatility calibrations—volatility, jump-risk, relative-value, and premium-repricing trajectories—into structured entry and exit decisions.
ML-backed trading systems
- Deployed Bayesian volatility-arbitrage options strategies using anomaly signals derived from stochastic-volatility calibrations, generating weekly returns of 5% on capital at risk.
- Built a regime-aware equity portfolio using Hidden Markov Models, Jensen–Shannon rebalancing, and NSGA-II Pareto-front optimization, producing 5% annualized alpha versus the benchmark in a multi-year backtest.
- Built an interday options platform using LoRA-tuned Chronos embeddings and per-asset TabPFN classifiers, with scheduled live execution across 18 equity and ETF underlyings.
Agents and infrastructure
- Built an MCP-compatible signal-mining agent that orchestrated hypothesis generation, analysis, and backtesting, surfacing strategies with Sharpe ratios above 1.5.
- Built a multimodal PDF research assistant using contrastive CLIP tuning, Qdrant retrieval, a two-pass vision-LLM workflow, FastAPI, and Next.js.
- Shipped containerized systems across AWS and Modal with automated deployment, DST-aware scheduling, persistent state, model caching, dashboards, and fail-closed execution recovery.
VISIE Inc. · Austin, TX
I joined during the early integration phase of a surgical navigation platform combining imaging and robotic actuation. My work centered on implementing TCP/UDP communication protocols for robotic arm motion control, supporting end-to-end product packaging and deployment, and helping stage a live demonstration that anchored the company's successful $8.2M Series A.
Sophelio · Austin, TX
I adapted physics-informed modeling originally developed for fusion experiment data to financial time series. The resulting production pipeline used sparse regression for PDE construction and signal generation, paired with a CAGR-maximizing Bayesian TPE optimizer driving a swing-trading system deployed on AWS Lambda.
Los Alamos National Lab & University of Southern California
I replaced 3D fracture geometry with a graph representation, running flow and transport as a reduced-order model. I also built sequential iterative coupling of flow and mechanics on different grids, tied together with a computational-geometry projection layer.
Education
Foundations
PhD, Engineering Mechanics
University of Texas at Austin
Austin, TX. Advanced training in numerical simulation, scientific computing, and mathematical modeling that continues to inform my work in ML and quantitative systems.
Life
Beyond the code
