Three years engineering production data platforms at Deloitte — GCP, BigQuery, Airflow, 1M+ records a day. Now pairing that foundation with RAG, multi-agent systems, and streaming. M.S. Business Analytics, UIC '26, 4.0 GPA.
End-to-end systems — not notebooks. Each one runs, deploys, and has a dashboard or API on the other end.
Multi-agent RAG platform built for TransUnion to automate Android security bulletin triage. A Sentinel agent classifies incoming CVEs into P0/P1/P2 risk buckets, a Coordinator agent routes and summarizes findings, and a Streamlit dashboard surfaces actionable intelligence. Deployed with GitHub Actions CI/CD and shared internally at TransUnion.
End-to-end real-time order streaming pipeline. A Python producer generates 2 orders/sec via Faker, Apache Kafka brokers the stream with Zookeeper coordination, a consumer inserts to Supabase, and a live Streamlit dashboard refreshes every 3 seconds with revenue, product, and regional metrics.
Production-style RAG assistant for UIC graduate program FAQs. Scrapy spider crawls program pages, OpenAI embeddings index content in Pinecone, FastAPI serves retrieval-augmented responses, and a Streamlit UI handles real-time Q&A. Reduces search friction for applicants navigating scattered program info.
Predicted loan defaults on a P2P lending platform using Decision Tree, Random Forest, and Gradient Boosting. Addressed class imbalance, evaluated with AUC/ROC and Lift curves, and ran investment simulations comparing model-based returns vs baseline options.
End-to-end churn prediction pipeline on 206k Instacart customers. Engineered 13 RFM-based behavioral features from 3M+ transactions, trained Logistic Regression and XGBoost classifiers, and flagged 27k high-risk customers across Low/Medium/High risk tiers. SHAP explainability shows which behaviors drive churn.
I'm looking for Data Engineering and Applied AI roles where pipelines meet production AI. Based in Chicago, open to relocating anywhere in the US.