Data Scientist

Implemented an image classification pipeline using CNN; collaborated on combining text and image model outputs via soft voting for a multi-modal product classification model. Designed and implemented a FastAPI-based MCQ generation API with HTTP Basic authentication, supporting filtered question retrieval and admin-restricted question addition. Developed an automated test framework for a sentiment analysis API using Docker Compose to orchestrate test containers for authentication, authorization, and response validation. Built an automated ML pipeline for data collection, preprocessing, and model training using Bash, cron jobs, and Make file.

Improved real-time recommenders on AWS, enhancing user engagement by filtering out previously consumed and sensitive content. Upgraded the autoplay recommender to handle more inputs, optimizing relevance and personalization. Implemented a workflow leveraging OpenAI LLMs on Databricks to automate learning outcome generation, covering high-volume metadata gaps.

Conducted customer behavior analyses to support product teams, including predictive modeling to identify user actions that signal future engagement. Supported product managers in KPI selection and A/B testing, and developed dashboards to monitor key metrics.

Carried out Fortran based numerical model experiments using parallel computing tools on a high-performance computing system. Explored the data using Python to quantitatively describe different flow conditions in the ocean.

Conducted disaster risk assessment and prepared analysis for various internationally collaborated research projects.