Skip to content
View BuffaloHat's full-sized avatar

Block or report BuffaloHat

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
BuffaloHat/README.md

Data & Analytics Leader

20+ years building the reporting, forecasting, and decision tools that help businesses grow

Bridging leadership, business strategy, and hands-on engineering — demonstrated here through the Soapy project suite, an end-to-end exploration of data engineering and applied LLM systems.


Featured GitHub Projects

Vinyl-collection visualization platform. End-to-end ELT: Python ingestion from the Discogs API → Parquet landing zone → dbt-core transformations on DuckDB (staging → intermediate → fact/dim marts), productionized with Prefect orchestration, PostgreSQL materialization, and containerized Metabase dashboards. Python · dbt · DuckDB · Prefect · PostgreSQL · Metabase

AI music-discovery tool over a ~10,000-track personal library. Dual-retrieval architecture pairing Essentia audio-feature vectors in Qdrant for acoustic similarity with natural-language RAG, served on Ollama (Gemma 27b) + Nomic embeddings, with a Streamlit interface for query, similarity search, and feature inspection. Qdrant · Essentia · Ollama · RAG · Streamlit

RAG research lab benchmarking four retrieval patterns — naïve vector, hybrid + reranking, GraphRAG, and an agent-maintained LLM wiki — against a curated 480+ article corpus with a rubric-scored evaluation harness. Pure Python with Qdrant, Neo4j, BM25, and cross-encoder reranking, governed by a Claude Code harness layer. GraphRAG · Neo4j · BM25 · Evals · Claude Code Harness

A 400+ page applied AI/LLM engineering reference for building production systems end to end — six parts spanning transformer mechanics, baseline & deployment decisions, RAG architecture, prompt & context engineering, harness design, agent memory & architectures, and evaluation / security / deployment ops. LLM Engineering · RAG · Agents · Evals · Production

Skills & Tools

Data Engineering & Analytics: Python · SQL · dbt · Jupyter · Prefect · DuckDB · PostgreSQL · Snowflake · Databricks · AWS · SAS · Git

BI & Visualization: MicroStrategy · Tableau · Metabase · BusinessObjects · Power BI · Alteryx

AI / LLM Systems: RAG · Hybrid RAG · GraphRAG · Agents · Multi-Agent Workflows · Prompt & Context Engineering · Rubric-Scored Evals

Retrieval, Vectors & Models: Qdrant · Pinecone · Chroma · Neo4j · BM25 · Cross-Encoder Reranking · Ollama · Nomic embeddings · Essentia audio features

Frameworks & Tooling: LangChain · LangGraph · LangSmith · Streamlit · Claude Code · Docker · Pandoc

Pinned Loading

  1. soapy-records-public soapy-records-public Public

    Personal vinyl collection project using Discogs API data. Tools: dbt, DuckDB, Postgres, Prefect, Metabase

    Python

  2. soapy-groovegraph-public soapy-groovegraph-public Public

    AI-enhanced music discovery using personal iTunes collection. Tools: Essentia, dbt, Qdrant, Ollama, Gemma, Streamlit

    Python

  3. soapy-raglab-public soapy-raglab-public Public

    Learning lab building 4 AI retrieval patterns - Karpathy wiki, naive RAG, hybrid+rerank RAG, GraphRAG - over AI research corpus with Eval

    Python

  4. soapy-ai-field-manual-search soapy-ai-field-manual-search Public

    A keyword-search application to view and query the 400+ page Soapy AI Field Manual

    Python