Pinned Loading

1. public-sprint public-sprint Public

   AI systems with a focus on NLP: retrieval‑augmented generation (RAG), agentic workflows, evaluation harnesses, and production patterns at individual scale. This repo is the entry point to my portfo…

   Jupyter Notebook

2. retail-copilot-gcp retail-copilot-gcp Public

   Architecture + scaffolding for a Retail Analytics Copilot (Local PoC) that turns natural language into validated SQL + VizSpec JSON over DuckDB. This repo mirrors the structure of a challenge dossi…

   Python

3. rag-light-demo rag-light-demo Public

   A local-first RAG demo that emphasizes production-grade patterns: deterministic SHA256 IDs for idempotent ingestion, hybrid retrieval (Chroma + BM25) fused with RRF, HMAC-signed BM25 indices, input…

   Python

4. agentic-reviewer-light-demo agentic-reviewer-light-demo Public

   LLM-powered semantic auditing for text classification. Clone, run, evaluate — no API keys, no cloud dependencies, everything runs locally with Ollama.

   Python

5. PCC PCC Public

   PCC Ecosystem: Privacy intent classification pipeline with 584-dimensional combined MiniLM and TF-IDF embeddings, dynamic GCS model ingestion, and BigQuery orchestration. Part of a production ML ec…

   Python

6. nanollama nanollama Public

   A rigorous, glass-box implementation of a modern Llama-style Transformer (RoPE, RMSNorm, SwiGLU) built from scratch for educational analysis and interpretability.

   Python