I'm Sneha

Designed and deployed a full analytics pipeline that mimics Netflix-style personalization. The system automated ingestion from multiple sources, trained a KNN recommendation model, and integrated LLM feedback loops to refine user suggestions in real time. Orchestrated with Prefect and deployed on GCP, the project demonstrates how combining robust data engineering with AI/ML can enhance personalization, engagement, and churn reduction.

Built an optimization model for hospital meal planning that balances cost efficiency, nutrition requirements, and dietary restrictions for 50+ residents. By formulating objectives and constraints, the model surfaced trade-offs between minimal cost and practical, diverse meal plans. The work highlights how operations research and optimization modeling can address complex real-world challenges in healthcare logistics.

Analyzed over 29 million U.S. flight records (2018–2022) to uncover disruption patterns across airlines, geographies, and seasons. Leveraged SQL queries for large-scale data wrangling, visual analytics (Tableau) for storytelling, and ML classifiers to predict delay durations. This project blends big-data storytelling and applied ML to improve operational efficiency, scheduling, and the passenger experience.

Developed an NLP pipeline to detect fraudulent job postings in ~18K listings. Combined Word2Vec embeddings with SMOTE balancing and ensemble models, achieving ~80% balanced accuracy. Analysis surfaced fraud signals such as vague job titles or missing company logos. The solution provides a scalable framework for job boards and recruiting platforms to improve trust, reliability, and candidate safety.