I build agentic AI systems — RAG pipelines, multi-agent frameworks, and the developer tooling that keeps them running in production.
- 🤖 Agentic AI — multi-agent orchestration with LangGraph, CrewAI, OpenAI Agents SDK & Claude Agent SDK; tool use, MCP, persistent memory, sub-agents
- 📚 Retrieval-Augmented Generation — production RAG APIs, vector databases, document parsing with provenance, evaluations & observability
- 🧠 LLMs & VLMs — prompt engineering, transformers, fine-tuning workflows, local inference with Ollama
- 🛠️ Developer tools — CLIs that clean up, audit, and automate the boring parts of a dev machine
- ⚙️ DevOps & infra — Docker, Terraform, Jenkins, CI/CD pipelines
| Project | What it does |
|---|---|
| Deep-Agents | Extensible AI agent framework — sub-agents, filesystem & shell tools, persistent memory, MCP support, and a terminal coding assistant |
| agentic-rag | Production-ready agentic RAG retrieval system with a RESTful API |
| agentcontext | Document parsing that never loses provenance — every block knows its source page, section, and location |
| diskteddy | Analyze disk usage and safely reclaim space from Docker, package caches & other space hogs — dry-run by default 🧸 |
| devteddy | Safely find, archive, and remove stale dev environments, dependencies, caches, and build artifacts |
| Smart-ticket-triage-agent | Agent that reads support requests, classifies issues, fetches order data, files Jira tickets, and notifies Slack |
| AI-Doctor-Assistant | FastAPI + React app demonstrating prompt management, tracing, evals, A/B testing, RAG & cost tracking in production |
| repodiff-auditor | Local-first web app for comparing Git repos & versions with evidence-based Q&A and downloadable reports |
Agent frameworks & SDKs
Models & inference
Vector databases & retrieval
💬 Open to collaborating on agentic AI, RAG systems, and developer tooling — connect on LinkedIn or reach me at harish12.21.04@gmail.com