Projects I've worked on

AeroDocs
Built a Next.js–based knowledge base for Honeywell to streamline data management using a retrieval-augmented generation pipeline for vector search. Implemented a powerful search system leveraging barcode scanning, part numbers, and image recognition with llava and yolo v8 models.

RAG & Vector Search

AutoChords
Developed an AI-powered platform that generates personalized chords based on a singer’s voice. built the backend with FastAPI and a custom voice recognition model using Fourier transform, with Next.js and TailwindCSS powering the frontend.

sunhacks' 23 winning project

FocusDucky
Developed an electron app that helps users stay focused by monitoring their screen activity. Utilized a locally run jina-embeddings-v3 model to detect distractions from natural descriptions.


gdsc hacks winning project

Schrodinger's AI
Worked on improving the integration of LLMs in government agencies like DHS by combining semantic entropy, RAG, and output linting to enhance data security, reduce misinformation, and increase transparency.

RAG & Semantic Entropy

RAG & Semantic Entropy

NivaChat
An AI assistant using sentence transformers, pytorch, and azure openai for semantic search, sop summarization, and intelligent follow-ups. i built a ticket routing engine combining embedding-based similarity (cosine similarity) with rule-based catalog mapping, along with nlp-driven metadata autofill, and integrated backend logic into the frappe framework with api-driven workflows for real-time ticket creation.

Book a call, we can talk about something cool!

© 2025 All right reserved

Made in Framer

Created by Bhoomi Sahajsinghani

Book a call, we can talk about something cool!

© 2025 All right reserved

Made in Framer

Created by Bhoomi Sahajsinghani

Book a call, we can talk about something cool!

© 2025 All right reserved

Made in Framer

Created by Bhoomi Sahajsinghani

Create a free website with Framer, the website builder loved by startups, designers and agencies.