Building systems at the intersection of machine learning, data pipelines, and AWS Cloud.
Solar-powered AIoT system for precision agriculture in monsoon-dependent farming regions. SVR-based rainfall prediction, ResNet50 crop identification, and a rule-based irrigation controller that only triggers when soil moisture, rainfall forecast, and crop health all agree.
Production-grade RAG pipeline combining dense and sparse retrieval over local PDFs. Reciprocal Rank Fusion + cross-encoder reranking, with automated RAGAS evaluation using synthetic QA generation.
Freelancing platform with a BERT + FAISS recommendation engine — 22-point precision improvement over TF-IDF across 10,000 profiles and 7,000 project records.
Emotion prediction system on a BiLSTM model, with activity recommendations, a Gemini API conversational layer, and a personal diary feature — deployed via Streamlit.
view repository →Hands-on AWS deployments on the free tier — static site via CloudFront CDN and an ML inference API on EC2. Learning cloud by shipping, not reading docs.
view repository →Lessons from designing a rainfall-prediction and irrigation-control pipeline for monsoon-dependent farms — model choices, sensor failures, and what actually shipped versus what stayed on the whiteboard.
Read full post →Data engineering pipelines, ingestion, transformation and analytics on AWS — covering S3, Glue, Redshift, Kinesis, and Athena.
Foundations of generative AI, large language models, prompt engineering, and responsible AI development on Google Cloud.
Machine learning workflows including model training, evaluation, deployment and MLOps practices using Google Cloud AI tools.
Structured approach to data science problems — from business understanding and data collection through modeling, evaluation, and deployment.
KLE Institute of Technology, Hubballi · Accepted for presentation & publication
Intellectual Property India · Government of India
Copyright Office · Government of India
University of Mumbai · Honours in AI/ML
A.P. Shah Institute of Technology