I’ve spent the last few years solving real-world problems across finance, healthcare, and AI research—not by throwing fancy models at them, but by building systems that work, explain themselves, and create actual impact.
At Cognizant, I helped financial teams screen 45,000+ loan applications using credit risk models I developed from the ground up—models that cut manual effort and improved recall on high-risk cases. At Recoup Health, I worked with clinicians to classify patient recovery outcomes based on wearable and survey data. Our model hit 96% balanced accuracy and directly supported treatment plans for 1,500+ remote patients. And at ASU, I built deep learning pipelines on regulatory filings (8,000+ S-1s) to predict IPO success—with interpretability baked in using SEC-BERT and SHAP.
Because data should mean something.
I care about fairness. About transparency. About building things that aren't just technically good, but actually usable by real people. That’s why I design models that are interpretable, pipelines that scale, and dashboards that make sense at a glance. Whether it’s a business team, a clinician, or a researcher—I want to hand them insights they can trust.
- A strong analytics and modeling foundation from my M.S. in Data Science at ASU (GPA: 4.0 / 4.0 With Distinction)
- 2 years of full-time industry experience at Cognizant in risk analytics and automation
- Real-world ML delivery experience in healthcare and financial domains
- Strong grasp of end-to-end pipelines—data engineering, feature work, model tuning, evaluation, and communication
- The ability to explain complex models clearly to non-technical stakeholders
- Regulatory Filing Classifier: Trained a model on 8,000+ SEC S-1 filings + financial ratios to predict IPO underpricing with 85% test accuracy
- Remote Health Risk Predictor: Used imbalanced data and time-series features from wearables to forecast patient recovery outcomes
- Vector Search Chatbot: Built a document-aware Q&A assistant using LangChain + FAISS over PDFs with 87% retrieval accuracy
- Time Series Forecasting: Applied ARIMA and Prophet to energy consumption data, reducing MAE by 15% and supporting grid planning
- 🥇 1st Place (50+ teams), Zoom AI Challenge — $4,000 Prize
Built a jargon-busting real-time assistant and professor-facing analytics dashboard - 🥈 2nd Place (70+ teams), Zoom Campus Spark Challenge — $2,500 Prize
Designed an ASL + multilingual accessibility layer for Zoom, aligning with inclusive design goals
- StrataScratch (450+ SQL problems solved)
Solved a wide range of case-based challenges involving joins, aggregations, filtering logic, subqueries, and business-ready analytics queries.
👉 View my StrataScratch profile
If you're building something that needs clarity, precision, and someone who will care deeply about the outcome—I'm open to opportunities in data science, analytics, or ML engineering.