Skip to content
View SebAustin's full-sized avatar

Block or report SebAustin

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
SebAustin/README.md

Sebastien Henry · AI & BI Engineer

I build AI systems and the data infrastructure they run on.

Applied AI Engineer at Action Company · Tableau Ambassador 2025 · Austin TX


🔧 How I Work

  • Traceability first. Every number should reconcile to a source of truth. If it can't, the feature isn't done.
  • Evals gate AI features. No eval set, no ship. Free-text interfaces without a measurable target are how AI products quietly break.
  • Cost is a design constraint. Warehouse sizing, caching, and query shape are part of the architecture, not a cleanup task.

🚀 Featured Projects

These are personal projects, built on public data. They're where I test patterns before I trust them anywhere else.

SEC EDGAR · FRED · Polygon.io · OAuth 2.1 · Claude Citations API · GitHub Actions

Production MCP server with citation-grounded 10-K extraction, deterministic eval harness (1.0 citation coverage), OAuth 2.1 + PKCE, and nightly CI/CD regression gates at 85%+ test coverage.



LangGraph · LangChain · FAISS · OpenAI · MCP

Multi-agent customer support system using LangGraph Supervisor Pattern — 4 specialized agents, RAG pipeline with FAISS vector store, MCP adapter layer, full test coverage.


PydanticAI · EDGAR-RAG · SQL · Regulation M

6-tool agentic EDGAR-RAG system coordinating SEC 10-K analysis, real-time market data, Fibonacci analysis, sentiment scoring, and Regulation M compliance checks.


Python · QLoRA · GRPO · DPO · vLLM · TRL

8-phase reproducible ML research pipeline on a 30B MoE model — rsLoRA, model soup averaging, custom GRPO reward functions, silent training/inference divergence fix. 4× L4 GPUs on Kaggle.


🛠️ Stack

AI/ML        │ LangGraph · LangChain · LangSmith · Claude API · OpenAI · AWS Bedrock
             │ QLoRA · LoRA · GRPO · Unsloth · vLLM · Fine-tuning
─────────────┼──────────────────────────────────────────────────────────────────────
Data         │ Snowflake · dbt · Spark · Glue · S3 · Lambda · Step Functions
             │ Tableau Cloud · Tableau Next · Pulse · REST API
─────────────┼──────────────────────────────────────────────────────────────────────
Engineering  │ Python · FastAPI · Pydantic · asyncio · n8n · Docker
             │ GitHub Actions · CI/CD · pytest · mypy
─────────────┼──────────────────────────────────────────────────────────────────────
Cloud        │ AWS (Developer Associate) · Azure · GCP

🏆 Recognition

  • 🥇 Indeed AI Innovation Hackathon Winner — Legal Operations Intelligence Engine (2025)
  • 🎯 Nominated · AI Recognition Grand Prize — Indeed (2026)
  • 🌟 Tableau Ambassador 2025 — Salesforce
  • 🎓 AWS Certified Developer – Associate

📬 Let's Connect

LinkedIn Email

Pinned Loading

  1. Autonomous-Knowledge-Agent Autonomous-Knowledge-Agent Public

    Multi-agent customer-support automation with LangGraph — RAG, tool execution, and human escalation

    Python

  2. mcp-financial-data mcp-financial-data Public

    MCP server (2025-11-25) for SEC EDGAR + FRED + Polygon with OAuth 2.1, citation-grounded 10-K extractor, and MCP Apps UI — Anthropic FDE / quant audience.

    Python

  3. multi-agent-trading multi-agent-trading Public

    Multi-agent stock analysis system with PydanticAI and Streamlit

    Python

  4. NVIDIA-Nemotron-Model-Reasoning-Challenge NVIDIA-Nemotron-Model-Reasoning-Challenge Public

    Python