Harshit Diwan

I’m an AI Software Developer who turns data into deployed systems that support real decisions—fast APIs, reliable pipelines, and measurable outcomes. I focus on building end-to-end solutions, from ingesting and transforming data to deploying AI services in Dockerized environments.

At Digitaaztrans Technologies, I developed and deployed a reverse geocoding system with an ETL pipeline moving large CSV datasets from AWS S3 into PostgreSQL, including geo-partitioning for optimized spatial queries. I also built a Flask REST API that supports proximity matching (exact or within 100 meters) and low-latency responses using Redis caching.

I’ve delivered multiple AI applications, including a RAG-based AI Investigator for the Chhattisgarh State Police using Ollama, Weaviate, and LangChain to identify entity relationships across confidential documents. I also built AI-powered monitoring and automation systems—an IoT water quality alerting workflow with automated notifications and reporting, and procurement facial recognition using Amazon Rekognition.

I’m comfortable working across computer vision and ML: I engineered Gemini 2.5 Flash Lite-based grain quality assessment and built a vehicle license plate recognition API with anti-spoofing detection. I also set up security monitoring infrastructure with Wazuh in Docker, and earlier experience includes analyzing sales data with Python/SQL and building Tableau dashboards.