Research & Open Source
Current Explorations:
- Graph Analytics & Anti-Corruption: Mapping global corporate ownership patterns with Neo4j.
- Prefect Orchestration: Robust patterns for production ML pipelines.
- Agentic Evaluation Frameworks: Strategies for governing autonomous AI workflows in production.
Bayesian Blocks
Jupyter
Python implementation of the Bayesian Blocks algorithm developed by Scargle et al. 2013 for time-domain astronomy.
Interactive i-RAG Quizmaster
Python
Inverting the traditional RAG paradigm by transforming dense, complex documents into a live-graded interactive examination engine built entirely in Python. Leveraging Docling for layout-aware chunking and DeepEval for synthetic dataset generation, rubric-based LLM-as-judge pipeline hosted in a reactive Marimo UI.
PyData MCP Workshop
Python
Hands-on workshop materials exploring Model Context Protocol (MCP) architectures. Covers building custom client-server tools and integrating them into modular, agentic LLM workflows.
PyData Prefect Workshop
Python
Materials for the PyData London 2024 Prefect Workshop. Covers building resilient data pipelines and workflow observability.