Atharva Patil

Machine Learning Researcher @Kodak Alaris

Rochester, NY, US

Engineered a multimodal event understanding pipeline integrating BLIP-2 for image captioning, YOLOv8 for object detection, Places365 for scene recognition, InsightFace for face/gender/age analysis, and FER+ for emotion detection.· Designed and implemented clothing/formality classification and enriched-context construction to enhance event inference from raw photos.· Built a zero-shot event classification framework (BART-MNLI) leveraging enriched context to identify 50+ event types (e.g, weddings, concerts, professional gatherings).· Optimized performance across modules with PyTorch, CUDA, TensorRT, and ONNX conversion, improving both inference speed and accuracy on real-world photo datasets.· Delivered reproducible, end-to-end ML workflows combining data preprocessing, feature fusion, and event classification into a scalable pipeline for applied vision–language AI.

Sanjay Ghodawat Institute

Grade 12th

Rochester Institute of Technology

Master's degree, Data Science

Savitribai Phule Pune University

Bachelor of Engineering - BE, Artificial Intelligence and Machine Learning