Contributed 80% to the development of tools to assess engagement in autismtherapy sessions.Created an end-to-end pipeline to process therapy videos and generate engagementscores.Implemented speech processing techniques to identify speakers, enhancing speakerclassification accuracy by 50%.Parsed therapy session reports to extract key information and generate insights usinglarge language models (LLMs).

•Specialized in working with geospatial data, employingtools such as Geopandas and Rasterio for efficient datamanipulation and analysis.• End-to-end development of machine learningmodels, encompassing data collection, featureengineering, and model architecture design.• Applied PyTorch for model training, ensuring robust andaccurate predictions in the geospatial domain.• Gained hands-on experience with MLops (MachineLearning Operations) by implementing MLFlow.• Streamlined the machine learning workflow, improvingmodel deployment, tracking, and management usingMLFlow.