I'm Harikesh Kushwaha, an alumnus of IIT Delhi and a Full Stack Gen AI Engineer currently at KPMG. I specialize in architecting and deploying production-grade agentic systems, enterprise RAG platforms, and automated data analysis tools with a focus on open-source, on-premise solutions for high-stakes government ministries and major financial institutions.
My work spans multi-agent orchestration (LangGraph, smolagents), advanced document understanding (OCR, hierarchical clustering), high-performance backend APIs, and polished full-stack UIs. I'm deeply interested in LLM fine-tuning, secure execution environments, and turning cutting-edge AI research into robust, impactful products.
I believe in continuous learning, open collaboration, and clean engineering. Feel free to explore my work or reach out. I'm always up for discussing AI, LLMs, or interesting engineering challenges.
- Agentic AI Orchestration — Multi-agent systems using LangGraph, smolagents, and DeepAgents with self-correction, ReAct loops, and supervisor–subagent patterns.
- Enterprise RAG — High-precision retrieval with hybrid search (Qdrant + Meilisearch, Semantic + BM25 + RRF), multi-hop reasoning, and hierarchical document clustering.
- Secure Execution — MicroVM-isolated sandboxes (Microsandbox) for safe agent-generated code execution in zero-trust environments.
- Gov Tech & Scale — National-level AI solutions supporting 22 Indic languages across central and state government ministries.
- Full Stack Development — End-to-end ownership: FastAPI backends, React/Vite frontends, real-time SSE streaming, and visual debuggers.
- Data Engineering & Analytics — Deep data pipelines integrating GenAI for automated visualization, statistical reporting, and schema-aware NL-to-SQL.
📈 GitHub Statistics