About Me

I build agent systems for industries where someone has to answer for them, and I get teams using them.

Agents that work. Teams that use them.

Years of quantitative modeling in regulated environments taught me what production-ready actually means: auditable, explainable, and built to survive scrutiny. I bring that discipline to AI agents. My most recent project, the Agent Accountability Harness, lets the person who deploys an agent declare what it may touch, record what it did, and report how each of its actions can be undone.

💼 Focus Areas

  • Agentic AI: Multi-agent systems with LangGraph, tool use through MCP, agent memory, and audit trails that record what an agent was allowed to do and what it did.
  • AI Enablement: Getting teams to use AI well, through teaching, demos, and tools people can run themselves.
  • AI Governance in Practice: Scopes, controls, and evidence that work at runtime, mapped to the NIST AI RMF, the EU AI Act, and the OWASP Top 10 for LLM Applications.
  • AI Orchestration & Multi-Agent Systems: LangGraph workflows, agent memory, dynamic routing, and audit trails for production deployment.
  • Model Development: Forecasting, scenario analysis, and credit risk modeling using Python, SAS, and SQL in production-level environments.
  • MLOps & Cloud Deployment: End-to-end pipelines with Docker, MLflow, FastAPI, AWS, and GCP.
  • RAG & LLM Engineering: Zero-hallucination architectures, schema validation, retrieval pipelines, and source-grounded generation..
  • Scalable Data Workflows: Developing modular, reproducible analytics using PySpark, partitioned data lakes, and serverless/cloud-native deployment for monitoring and performance.

🎓 Teaching

I am a lecturer at the University of Texas at Dallas, Naveen Jindal School of Management, where I teach BUAN 6V99, Agentic AI & Process Automation.

📚 Academic & Research Background

Educated in the U.S. and U.K., I’ve published peer-reviewed research on foreign investment and economic policy. My academic rigor informs my approach to real-world data challenges. I believe the greatest value lies not in a perfect model, but in a model you can trust, explain, and maintain. Beyond data, I’m drawn to storytelling, visual communication, and exploring how complex ideas become intuitive.

🌍 Languages & Interests

Fluent in English, Spanish, and Portuguese; advanced in French and Italian. I enjoy exploring languages, data visualization, and refining communication through storytelling.

🔗 Connect

📬 Newsletter: Exit Code 2 | AI Under Audit
Governance that works. Guardrails that execute. Subscribe on LinkedIn Subscribe on Substack (free)

🛠️ Core Tools & Skills

AI & Agent Systems: LangGraph • LangChain • Model Context Protocol (MCP) • Claude API • Claude Code • OpenAI API • Retrieval-Augmented Generation (RAG) • Embeddings & Vector Stores (Chroma, FAISS)

Languages & ML:
Python • scikit-learn • SQL • PySpark • XGBoost • Prophet • SHAP • R • Stata • SAS

Cloud & MLOps:
AWS (S3, SageMaker, Lambda) • GCP (Cloud Run, BigQuery) • MLflow • Render • Docker • GitHub Actions (CI/CD)

Data & Workflow:
Parquet • DuckDB • Neo4j • Pipeline design (Bronze → Silver → Gold)

Visualization & Apps:
Streamlit • Tableau • Plotly • Matplotlib

Other:
GIS (Certified)