Senior AI/LLM engineer and founding engineer focused on production AI systems, scalable Django/Python backends, RAG pipelines, Langfuse evaluation workflows, Strands agents, Pinecone/Cohere retrieval, and cloud infrastructure for healthcare software.

Currently building backend and AI systems for Stella at Ourself Health.

alexbeattie.com · AI Engineer · Projects · LinkedIn

• Production AI systems: routing, tool selection, safety gates, source metadata, Strands agent behavior, and observable decision flows
• RAG and retrieval: pgvector, Pinecone, Cohere embeddings, source metadata, query rewriting, grounding, and retrieval evaluation
• LLM evaluation: Langfuse traces, gold-case evals, LLM-as-judge workflows, deterministic retrieval metrics, and live trace scoring
• Model economics: cost-aware routing, token analytics, budget alerts, and smaller-model paths for simple queries
• Backend architecture: Django, Python, GraphQL, PostgreSQL/PostGIS, Redis, and API contracts
• Cloud infrastructure: AWS Bedrock, ECS, Aurora, S3, CloudFront, GitHub Actions, and cost optimization
• Mobile-connected systems: Flutter, SwiftUI, streaming APIs, auth flows, and mobile/backend contracts
• AI development workflows: Cursor custom rules/skills, Claude Code, MCP servers, and IDE-native engineering context

Founder/engineer case study for a nonprofit startup platform helping families find developmental disability services across Los Angeles County. The project started as a commissioned web mapping system and grew into a live web platform, iOS app, Django/PostGIS backend, Bedrock AI assistance, and a Cohere/Pinecone research RAG stack.

Stack: Python, Django, Django REST Framework, PostgreSQL, PostGIS, Vue, Mapbox, SwiftUI, AWS Bedrock, Strands, Cohere, Pinecone

Backend and AI architecture for Stella at Ourself Health: a production health assistant with source-grounded RAG, streaming responses, clinical knowledge retrieval, Langfuse observability, evaluation workflows, and mobile/backend contracts.

Stack: Python, Django, Strawberry GraphQL, AWS Bedrock, pgvector, Aurora PostgreSQL, ECS Fargate, Langfuse, Auth0, Flutter

An installable Python package and MCP server that lets AI agents work across GitHub, Slack, Jira, Confluence, Google Docs, and Miro from one consistent action layer. Includes a CLI, provider connectors, OAuth refresh handling, webhook helpers, and tests.

Stack: Python, Model Context Protocol, FastMCP, OAuth 2.0, GitHub API, Slack API, Atlassian Cloud, Google Docs/Drive API, Miro API

• Designing Production AI Routing And Evals For A Healthcare Assistant
• Lessons From Bedrock, pgvector, And RAG In Production
• Streaming LLM Architecture Patterns: Sources, Done Events, And Observability

AI/Agents: RAG pipelines, MCP servers, multi-agent orchestration, LLM-as-judge evals, model routing, vector search, AWS Bedrock, Claude, Cohere, Pinecone

Backend: Python, Django, GraphQL, REST APIs, PostgreSQL, PostGIS, Redis

Cloud: AWS ECS, Lambda, Bedrock, Aurora, S3, CloudFront, Docker, GitHub Actions

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