aksh modi

I’m a research-oriented data science student with a strong foundation in developing and deploying scalable AI/ML solutions. I’m passionate about leveraging cutting-edge technologies across NLP, computer vision, and recommendation systems.

I’ve built end-to-end models and systems in Python using TensorFlow, with hands-on experience in model monitoring and evaluation pipelines. On the engineering side, I’ve deployed tools and services that track performance reliably, including an automated ML model monitoring pipeline that improved reliability by 25%.

My projects reflect my focus on real-world impact: AquaTrace uses custom-trained YOLO models on underwater imagery to predict coral bleaching up to 10 days ahead and deliver insights through a WhatsApp-first interface and dashboard. I also implemented a Retrieval-Augmented Generation (RAG) pipeline for daily financial intelligence reports, including data ingestion, semantic indexing, and vector-based retrieval to improve contextual relevance.

I’m also actively contributing through open-source work in the Wikimedia ecosystem, collaborating on MediaWiki development and tool maintenance. My research progress includes a published patent for a soil sensor monitoring system and accepted conference work on lightweight YOLOv8s coral bleaching classification and swarm-optimized deep learning for agricultural IoT failure detection and prediction.