Skip to content
View alexbeattie's full-sized avatar

Block or report alexbeattie

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 supported. This note will be visible to only you.
Report abuse

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

Report abuse
alexbeattie/README.md

Alex Beattie

AI/LLM Engineer building retrieval-augmented generation systems, multi-agent orchestration, and production AI applications. Focused on practical, cost-efficient AI that runs locally or on AWS.

Blog | LinkedIn

AI & LLM Projects

Fully offline RAG pipeline that indexes an Obsidian vault into ChromaDB using heading-based semantic chunking and nomic-embed-text embeddings via Ollama. Includes an MCP server so coding agents (Cursor, Claude Code) can query the vault during development. Interactive REPL with streaming LLM responses and source citations.

Stack: Python, ChromaDB, Ollama, FastMCP, nomic-embed-text, mistral-nemo

Lightweight evaluation framework for RAG pipelines. Three layers: retrieval metrics (recall@k, precision@k, MRR), LLM-as-Judge response scoring (relevance, groundedness, completeness), and claim-level hallucination detection. ~300 lines, zero framework dependencies.

Stack: Python, Ollama, LLM-as-Judge pattern

Intelligent query routing that classifies complexity with zero-latency heuristics and directs simple queries to a fast model, complex queries to a capable model. Reduces inference costs ~70% while maintaining quality. 100% accuracy on built-in benchmark.

Stack: Python, Ollama, regex-based classification

Production Apps

Healthcare provider mapping app with Django/PostGIS backend and native SwiftUI iOS app. AI-powered search using AWS Bedrock with Strands SDK. Live on the App Store.

Stack: Python, Django, PostGIS, AWS Bedrock, SwiftUI, iOS

Real-time GPS vehicle tracking app with Vue.js frontend and Go backend.

Stack: Vue.js, Go, REST APIs

What I Work With

AI/ML: RAG pipelines, multi-agent orchestration, LLM-as-Judge evaluation, model routing, vector databases (ChromaDB, pgvector), embedding models, MCP servers, Ollama, AWS Bedrock

Backend: Python, Django, Go, Node.js, GraphQL, REST APIs

Cloud: AWS (ECS, Lambda, Bedrock, Aurora), Docker, CI/CD

Frontend & Mobile: Vue.js, React, SwiftUI, Flutter

Pinned Loading

  1. CHLA CHLA Public

    Django + PostGIS backend and SwiftUI iOS app for healthcare provider mapping (live on App Store)

    Python 2

  2. obsidian-rag obsidian-rag Public

    Local RAG pipeline for Obsidian vaults using ChromaDB + Ollama (nomic-embed-text)

    Python

  3. riottracker riottracker Public

    Vue.js real-time tracking dashboard

    Vue

  4. aurora-postgresql-pgvector aurora-postgresql-pgvector Public archive

    Forked from aws-samples/aurora-postgresql-pgvector

    Agentic AI Use Cases with pgvector, Aurora PostgreSQL and Amazon Bedrock

    Python

  5. llm-eval-harness llm-eval-harness Public

    Lightweight eval framework for RAG pipelines: retrieval metrics, LLM-as-Judge scoring, hallucination detection

    Python

  6. model-router model-router Public

    Intelligent query routing: classify complexity with zero-latency heuristics, route to fast or capable LLM models

    Python