Kaushik Kumar Giri

I'm a Full Stack AI Developer with over 2 years of experience at HanuAI, where I've built production-grade AI-powered web applications end-to-end, from backend architecture to deployed features used at government scale.

My core strength lies in building scalable backends with Django, FastAPI, and Django REST Framework, paired with React.js and React Native frontends. I've optimized PostgreSQL schemas and queries on large datasets (cutting response times up to 30%), built asynchronous processing pipelines with Celery and Redis that reduced API latency by 40%, and implemented Redis caching and throttling that cut high-traffic latency by 35%. I've also worked extensively with PostGIS and GeoServer for spatial querying and map visualization on geospatial datasets.

What I'm most passionate about is the AI/GenAI side of my work. I've designed and shipped a LangGraph-based agentic workflow with tool and function calling for automated report generation, and orchestrated a multi-agent validation pipeline using LangFlow to review computer-vision detections and improve output reliability. I fine-tuned a YOLOv8 computer-vision model and an open-source LLM on custom road-defect datasets, which cut manual review effort by roughly 70% in a real production pipeline for automated pothole and signage detection.

One of my favorite projects has been building an AI SQL-assistant chatbot: a full RAG pipeline (LangChain + pgvector) that lets non-technical users ask natural-language questions and get back SQL-driven answers with auto-generated chart visualizations, grounded in domain documents through semantic search. I even added voice interaction (TTS/STT) so users could query infrastructure data hands-free. On the side, I built FaceAndFlare, a personal project combining a FastAPI backend, JWT/RBAC auth, and a Gemini + Pinecone RAG assistant for natural-language service recommendations, containerized and deployed on AWS.

I hold an MCA from NIT Calicut, and I'm generally drawn to roles where I can work across the full stack while going deep on applied AI: RAG systems, agentic workflows, LLM fine-tuning, and computer vision, especially in products that solve real infrastructure or operational problems at scale. I'm comfortable owning a feature from database design through deployment (Docker, AWS, CI/CD), and I enjoy environments where I get to work closely with frontend, mobile, and QA teams through the full SDLC.