Sagorika Ghosh

About Sagorika Ghosh

I’m an Applied Scientist with experience building ML and AI systems that drive measurable business impact. I’m currently pursuing my M.S. in Data Science at the University of Washington, graduating March 2026.Most recently, I worked as an Applied Scientist Intern at Uber, where I built and deployed predictive models and ran causal experiments to improve marketplace reliability at scale. My work spanned ETA and delivery-time modeling, robust statistical analysis, and A/B experimentation under real-world supply and demand constraints. Prior to that, I worked at American Express as a Data Scientist II, developing forecasting models, fraud detection systems, and production ML pipelines deployed across global markets.I’m currently working on a safety-critical, domain-specific RAG system at Contextual AI, focused on document ingestion, metadata-driven retrieval, grounded response generation, and evaluation. This includes building retrieval pipelines over regulatory and OEM documents, implementing citation-backed generation, and deploying an end-to-end WhatsApp chatbot with reliability and hallucination checks.Highlights:• Built Deep Set neural networks for delivery time prediction (R² = 0.97), improving ETA accuracy by 6%+ and reducing courier wait time• Designed and analyzed CUPED-based A/B experiments, reducing courier lateness by 2–3 minutes in undersupplied markets• Builing a safety-critical RAG system ingesting 150+ regulatory and OEM documents, implementing metadata-filtered retrieval and citation-grounded generation.• Developed LSTM, CNN-LSTM, and BiLSTM forecasting models, improving prediction accuracy by up to ~30%• Improved fraud detection models using XGBoost, increasing recall at high-risk thresholds and driving measurable cost savings• Built and deployed large-scale ETL pipelines and ML systems supporting production workloads across multiple global regionsI’m seeking full-time Applied Scientist or ML Engineer roles starting post March 2026, with a focus on applied machine learning, experimentation, ML systems, and LLM/RAG-based applications.

Experience

1. Aws Reachback Technical Cohort

   Amazon Web Services (AWS)

   Feb 2026 — Present · Seattle, WA, US

   Selected for the AWS Reachback Technical Cohort 2026, gaining hands-on training in generative AI workflows on AWS, including Amazon Bedrock-based prompting, RAG concepts and patterns, foundation model selection and customization, inference tuning (temperature, top-K, top-P), and responsible AI practices.

Education

• Arwachin International School - India

  Senior secondary (10+2), CBSE

• Guru Gobind Singh Indraprastha University

  Bachelor of Technology - BTech, CSE

• University of Washington

  Master of Science - MS, Data Science