Eric Holt

Featured Projects

TraceAgent — Observable Agentic Research Platform

Python · FastAPI · Groq · Tavily · React · PostgreSQL · Docker · AWS

Built around a four-stage pipeline — planner, search, reflection, and synthesis — where the reflection stage evaluates result adequacy and conditionally loops back to search (up to 3 times) with refined queries before report generation. Every internal decision is persisted to PostgreSQL and streamed to the browser in real time via WebSocket.

Deployed to Hugging Face Spaces (Docker) and AWS EC2 (t3.micro) + RDS PostgreSQL (db.t4g.micro, VPC-only) · CloudWatch Logs via watchtower · 6 custom CloudWatch Metrics per run (RunsCompleted, PipelineDurationMs, ReflectionLoops, and more) feeding a live pipeline dashboard · Public activity feed — browse all runs with full attribution · Fork any run with a modified query, lineage tracked · Side-by-side run comparison · Multi-stage build bundles React/Vite frontend with FastAPI backend

Live Demo · Code · README · AWS Demo

DocSignal — AI-Powered Documentation Intelligence

Python · FastAPI · React · Groq · pgvector · PostgreSQL · Docker

Ask natural language questions against any documentation set — DocSignal retrieves semantically relevant answers with cited source passages, not just keyword matches. Ships with preset doc sets (FastAPI, SQLite, Requests) and supports ingesting any PDF or URL (up to 25 pages) into a custom set.

fastembed BAAI/bge-small-en-v1.5 local embeddings (no API key) · pgvector cosine similarity search · Groq LLM answer synthesis with citations and similarity scores · Analytics dashboard: query frequency, top sources, volume over time · Multi-stage Docker build · Deployed to Hugging Face Spaces

Live Demo · Code · README

DailySignal — Automated AI News Briefing Service

Python · GitHub Actions · Groq · PostgreSQL · Neon · Resend · React · Recharts · Docker · Hugging Face Spaces

Twice-daily AI news briefing service — fully automated from fetch to inbox with zero manual steps. GitHub Actions cron pulls from HackerNews, Product Hunt, and five RSS feeds, deduplicates via SHA-256 hashing, synthesizes a structured briefing via Groq, and delivers it to subscribers via Resend.

GitHub Actions cron scheduling — no always-on server required · Cross-source deduplication using SHA-256 content hashes · Per-subscriber unsubscribe tokens with automatic deactivation after 3 failed deliveries · PostgreSQL audit log per subscriber per run · 55 unit tests, 98% coverage · Deployed to Hugging Face Spaces with Neon serverless PostgreSQL

Live Demo · Code · README

RepoSignal — GitHub Issues Analytics Dashboard

Python · FastAPI · Groq · PostgreSQL · React · Vite · Tailwind CSS · Recharts · Docker

Enter any public GitHub repo and RepoSignal fetches up to 50 issues via the GitHub REST API, classifies each one by type, priority, and sentiment using an LLM (parallelized with a concurrency cap to stay within Groq rate limits), persists results to PostgreSQL, and renders a live analytics dashboard. Select any open issue to stream an AI-drafted triage response word-by-word in real time.

SQL-powered dashboard: COUNT + GROUP BY, time-series grouping by week, JOIN across issues and classifications · Filterable issue table by type and priority · SSE-streamed AI triage drafts · Repo history — previously analyzed repos saved to DB and re-loadable from the landing page · 9 API routes with full unit + integration test coverage · Multi-stage Docker build · Deployed to Hugging Face Spaces

Live Demo · Code · README

Get Set Wedding — Production SaaS Planning Platform

React · TypeScript · Supabase · Stripe · Node

Built end-to-end as contract full-stack developer — database schema, PostgreSQL RLS for structural multi-tenant isolation, Stripe billing with webhook-driven server-side enforcement, and a context-aware AI assistant integrated with live user planning data.

PostgreSQL-layer RLS (structural isolation, not application-level) · Webhook-driven Stripe pipeline — server-side enforcement on every API request prevents client-side bypass · Refresh-safe JWT session rehydration · Demo-to-account conversion with isolated storage scopes · RBAC admin routing · AI assistant with live access to user planning data (budget, guest list, vendors)

Live App · README

TriageAgent — AI-Powered Email Triage & Draft System

Python · FastAPI · Groq · Gmail OAuth2 · React · Docker · GitHub Actions

Three-stage pipeline — Classify, Extract, Draft — processes pasted emails end-to-end: classifying type and urgency, extracting sender intent and action items, then generating a professional reply. Every stage streams structured output to the browser in real time, and results with confidence below 0.75 are flagged for careful human review before sending.

Urgency scoring (low / medium / high) with confidence threshold safety gate · Gmail OAuth2 for direct send from the UI · Groq llama-3.3-70b-versatile inference · Dockerized single-service deploy to Hugging Face Spaces · Automated CI/CD via GitHub Actions on push to main

Live Demo · Code · README

AgentMesh — Multi-Agent Code Review & Repair System

Python · LangGraph · FastAPI · Groq · React · Tailwind CSS · Docker

Three specialized agents collaborate via a LangGraph StateGraph — Reviewer identifies issues, Fixer produces corrected code with a changelog, and Evaluator scores the fix across three dimensions and can reject the output and route back to Fixer with specific feedback for a second pass. Every agent decision and handoff streams to the browser in real time via SSE.

Evaluator scoring: Correctness (40%) · Completeness (40%) · Code Quality (20%) — score 50–69 triggers Fixer retry with evaluator feedback, capped at 2 passes · Structured Pydantic schemas for all inter-agent state · File upload or JSON input, language auto-detection · Stateless backend · Multi-stage Docker build deployed to Hugging Face Spaces

Live Demo · Code · README

AgentFlow – AI-Powered Document Intelligence

Python · LangGraph · FastAPI · Groq · React · Docker

Built on a LangGraph multi-step agent pipeline that classifies uploaded HR documents (resumes, cover letters, interview notes, scorecards, job descriptions) and routes each to a specialized LLM analysis node returning structured JSON with skills, risk flags, and hiring recommendations.

Groq LLM inference (llama-3.3-70b) + tenacity retry · Vision model support for image/handwritten docs (llama-4-scout) · Deterministic classification pre-check · Single Dockerized service on Hugging Face Spaces · SQLite session persistence · n8n webhook integration

Live Demo · Code · README

Transportation-LLM — Local AI Chat via WebGPU

WebGPU · TypeScript · RAG · MLC-AI

Built around Llama-3-8B running entirely in-browser via WebGPU — no backend, no API calls. A domain-specific RAG engine using all-MiniLM-L6-v2 embeddings retrieves and ranks transportation knowledge (rail, automotive, transit, standards) before augmenting each prompt, with the semantic index pre-built offline and served statically.

WebGPU acceleration via @mlc-ai/web-llm · On-device inference with no server infrastructure · Semantic RAG with domain filtering, weighting, and diversification · Pre-built static index via @xenova/transformers · Python RAG index builder with configurable chunking · Deployed via GitHub Pages

Live Demo · Code · README

WarehouseBot-RL — Autonomous Picking Simulation

Python · stable-baselines3 · ONNX · Reinforcement Learning · Cloudflare Workers

Built around an RL agent trained offline with stable-baselines3 and exported to ONNX for in-browser execution. The bot plans pick sequences using nearest-neighbor and 2-opt heuristics, navigates obstacles via A* pathfinding, and accumulates rewards based on efficiency and safety metrics across three rotating warehouse layouts.

RL training via stable-baselines3 · ONNX export for browser compatibility · A* obstacle navigation · Nearest-neighbor + 2-opt pick sequencing · Cloudflare Workers + Durable Objects for persistent shared stats across all visitors · Canvas rendering on mobile and desktop

Live Demo · Code · README

GitHub Activity

About

I design and build full-stack AI systems focused on real-world deployment, observability, and scalable architecture, combining agent orchestration, semantic search, and cloud infrastructure.

My work spans multi-step agentic pipelines with conditional looping and self-evaluation, RAG systems with pgvector and cited retrieval, React/Vite frontends with live SSE and WebSocket streaming, production SaaS with Stripe billing and PostgreSQL RLS, and automated scheduled pipelines with email delivery across multi-source data aggregation. I instrument systems with custom CloudWatch metrics and live dashboards, and ship with CI/CD pipelines. Not just working prototypes, but production-grade services with monitoring and lineage.

Contact

Email · GitHub · LinkedIn