Sam

Hybrid ML + CrewAI system to triage millions of emails: fast TF-IDF+SVM gate, agentic workflows, CLI & API.

Descriptions: - **Hierarchical Clustering**: It's an unsupervised machine learning algorithm that builds a hierarchy of clusters by continuously merging the most similar pairs of data points until only one cluster remains. - **TimeSeriesKMeans Clustering**: This is a variation of KMeans clustering, specifically designed for time series data.

Deep document understanding for clinical reports — parse into a sections+tables tree, extract lab fields, and retrieve with section/row-level citations. Inspired by RAGFlow

Governed LLM gateway — PHI guardrails (redact/block), multi-provider routing with fallback, and per-call cost accounting. OpenAI-compatible, built on LiteLLM

Program-not-prompt clinical message router — a DSPy-style Signature/Module/Optimizer loop that measurably improves routing accuracy (baseline vs optimized)

Clinical GraphRAG — typed medical entity extraction, a co-occurrence knowledge graph, and dual-level (local/global) source-traceable retrieval. Inspired by LightRAG