Projects

Geosentinel AI (Vista Global Hackathon - Agentic AI Winner – 2025)

GeoSentinel AI is an intelligent dashboard designed to assess and visualize disaster-related risks for properties across the U.S., leveraging FEMA data and AI agent-driven analysis. The platform offers a national-level risk overview highlighting “attention needed” areas to kick-start deeper investigations.

It is an interactive agentic system to asses the roof quality and predict roof damage FEMA score for home owners by analyzing the history of roof damages captured in Eagleview.

By combining geospatial data, historical records, and machine learning - GeoSentinel AI empowers stakeholders to make data-driven decisions for disaster preparedness, mitigation, and recovery planning.

Stack: Autogen, AWS bedrock, LLM, VLM and Python

Impact: Peer Choice Award at Vista Global Hackathon and a fist ever win for Eagleview at Vista Hackathon

EVInsight (Hackathon Finalist – 2024)

A chatbot that uses aerial imagery and a Vision-Language Model to answer property-based questions interactively.

Stack: Mapbox, RAG, VLM, Python

Impact: Real-time answers for solar, insurance, and construction.

Edit your photo with AI - Inpainting

A weekend project using stable diffusion for Inpainting. We have made the user experience easy with automatic mask generation, Editing the unwanted objects on the picture. Easy editing.

Stack: GenAI, Stable Diffusion, Python, AWS

Impact: Anybody can edit the photo for free

Github

editurphotowithai-demo.mov

Virtual Inventory Management

Helps the store managers on store managers on update of the number of products on the shelf. Towards creating virtual inventory management for retail stores.

Stack: Deep learning, SSD, Pytorch, Python

Impact: Shelf Analytics, Manage inventory with Detection models

Github

Emotion Estimation

A fun side project on emotion estimation with Friends clip.

Stack: Face detection - MTCNN, Trained model using FER2013 data, Python, Deeplearning

Impact: Real-time emotion detection with face

Github

emot_est.mp4

covid19_actual_sd2.mp4

Ensuring Social Distancing among people via surveillance cameras to fight against Covid-19

To ensure social distance among people via surveillance cameras, used a monocular depth algorithm to get better proximity information. This algorithm can be used on any camera view.

Stack: Python, Deep learning, Depth estimation, Distance metrics

Impact: Real-time Distance awareness

Github

Page updated

Google Sites

Report abuse