Data Scientist, ML or AI Engineer

Built robust, reproducible pipelines to clean, enrich, and transform over 26,000 unstructured conflict event records from the ACLED dataset, enabling high-quality GenAI analysis. Applied advanced topic modeling techniques using LLaMA 2, BERTopic, and LangChain to extract latent protest themes, boosting topic coherence by 27% and topic diversity by 120%. Utilized AWS EC2 instances to run compute-intensive LLM workflows, and AWS S3 for efficient data management and sharing. Automated parts of the workflow using AWS Lambda to reduce manual overhead and streamline preprocessing. Delivered structured insights via Jupyter Notebooks and technical reports, improving stakeholder onboarding and decision support.

Cleaned and analyzed over 1 million financial and operational records from multiple data sources, ensuring 99% data accuracy for downstream analytics. Automated recurring tasks using Microsoft Power Automate, reducing manual workload by 30% and saving over 15 hours per week. Developed interactive Power BI dashboards and performed exploratory and statistical analysis using SQL and Python, enabling data-driven decisions and improving process efficiency by 20%.