Hi, I'm Avishek Saha.
I build applied AI systems for operational workflows, especially multimodal agents, tool-using LLM applications, and Python backends.
My strongest recent project is Maintenance-Eye, a real-time maintenance copilot built with Google ADK, Gemini Live API, FastAPI, Firestore, and Cloud Run. At BC Rapid Transit Company, I work on data pipelines and prototype maintenance-focused AI workflows.
Featured Projects
- Maintenance-Eye — A deployed multimodal maintenance copilot for field technicians, built with Google ADK, Gemini Live API, FastAPI, Firestore, and Cloud Run.
- Price Prediction Platform on AWS — An earlier end-to-end ML deployment project for house price prediction using Flask, AWS ECR/EC2, Docker, and GitHub Actions.
- Autism Spectrum Disorder Prediction — Classification models using SVM and deep learning (CNN) for ASD detection with 94% accuracy. Published in IJITCS.