Data Scientist

Trained an LSTM fake-news classifier on 50,000+ news articles, achieving 85%+ accuracy through targeted preprocessing, tokenization, and feature engineering. Shipped a GUI verification tool that lets non-technical users check article credibility in real-time, translating a research model into a usable product. Owned the full ML workflow from data ingestion through model evaluation, working in an Agile team with code review and versioning.

Analyzed 100,000+ rows of production data to surface throughput bottlenecks and process inefficiencies, feeding insights directly to plant supervisors. Built an interactive Looker Studio dashboard that lifted reporting efficiency by ~40% and gave management daily visibility into production KPIs. Replaced manual paper-based data collection with a Google Forms pipeline, cutting paperwork by ~60% and lifting data accuracy by ~30%.

Cleaned and modeled 1M+ healthcare records using Python and SQL, reducing inconsistencies by ~95% and producing analysis-ready datasets for downstream reporting. Built ETL workflows and predictive models that increased processing efficiency by ~40% versus the legacy pipeline. Designed Looker Studio dashboards communicating critical insights to non-technical stakeholders, improving stakeholder comprehension by ~30%.