ABOUT ME
I'm James Potter, a data engineer and technical developer based in the scenic mountains of Western North Carolina. For the past two years, I've been intentionally re-skilling into AI/ML infrastructure and developer tooling after previously working as a pharmacy technician. My work focuses on building documentation systems, developer tools, and data infrastructure that makes complex technical information accessible to both humans and AI agents.
What I do
I build developer tools and infrastructure for AI/ML workflows, specializing in documentation search systems, vector databases, and data processing pipelines. My current focus is on the Model Context Protocol (MCP) ecosystem, where I've created a framework that transforms technical documentation into intelligent, searchable resources for AI assistants. I work extensively with emerging technologies including Mojo programming language, Modular's MAX framework, and modern data tools like DuckDB and Observable Framework.
Recent Projects
- MCP Documentation Framework MCP
Built a production framework for converting markdown documentation into searchable MCP servers with hybrid vector + keyword search. Created self-contained servers for Mojo language documentation, DuckDB docs, and the MCP protocol itself. Implements semantic search using MAX embeddings (sentence-transformers) combined with full-text search via DuckDB's HNSW indexes and BM25 ranking.
- Hybrid Search Implementation
Designed and implemented Reciprocal Rank Fusion (RRF) algorithm combining vector similarity search with keyword matching for improved documentation retrieval. System includes graceful fallback to keyword-only search when vector embeddings are unavailable, demonstrating production-ready error handling.
- Multi-Language Programming Examples Go-Vols
Created beginner-friendly repository demonstrating "Hello World" equivalents across 10 programming languages (Python, Rust, Go, C, C++, Java, JavaScript, SQL, Bash, Fortran) with detailed setup documentation and CLI tutorials for newcomers to programming.
- Data Visualization Projects
Built interactive data visualizations using Observable Framework, including restaurant recommendation mapping with automated geocoding pipelines and sports statistics dashboards. Demonstrates proficiency in web-based data presentation and ETL workflows.
Technical Stack
Core Languages & Tools:
Python (primary), SQL, Mojo (learning), Markdown, HTML/CSS/JavaScript
Data & Infrastructure:
DuckDB, PostgreSQL, Ducklake, Vector Databases (HNSW indexes), Pandas
AI/ML Tools:
Modular MAX Framework, Sentence Transformers, MCP Protocol, Vector Embeddings
Developer Tools:
Git/GitHub, VS Code,Docker, UV/Pixi (Python Package Management), GitHub Actions
Frameworks:
Observable Framework, Scrapy, FastMCP, Marimo notebooks
Current Focus
About My Journey
After earning an associate's degree in bioscience technology, I worked as a pharmacy technician for several years. Two years ago, I made a deliberate decision to pivot into data engineering and developer tooling. This transition has been self-directed—I've learned by building production tools, contributing to open-source ecosystems, and staying current with emerging technologies in the AI/ML space.
I currently work part-time at a local hiking shop while pursuing full-time opportunities in technical writing, developer relations, or data engineering roles. My partner's career keeps us based in North Carolina, making remote positions particularly appealing, though I'm happy to travel for onboarding, training, or team meetings.
I'm available for freelance data projects and consulting. Whether you need help cleaning messy datasets, extracting information from documents, or building data pipelines, I'd love to discuss how I can help. Feel free to reach out through LinkedIn or explore my projects on GitHub.