Applied AI · Research · New York
I build applied AI systems that connect machine learning, product thinking, and real-world infrastructure: from recommender platforms and forecasting pipelines to responsible AI research and safety-critical healthcare systems.
Applied AI · Research · New York
I build applied AI systems that connect machine learning, product thinking, and real-world infrastructure: from recommender platforms and forecasting pipelines to responsible AI research and safety-critical healthcare systems.
I'm an Applied AI graduate student at Stevens Institute of Technology (MS, Applied Artificial Intelligence, Data Engineering concentration), building toward a career as an Applied ML Systems Engineer at the intersection of research and deployment.
My work focuses on trustworthy AI for safety-critical and resource-constrained systems, spanning healthcare, finance, and operations.
My background spans process engineering (BASc, University of British Columbia), data analytics, ML deployment, and now active research publication, giving me an unusual mix of scientific rigor, engineering discipline, and product thinking.
Languages: English (Professional Proficiency) · Hindi (Native)
Quick Facts
By the numbers
0
Peer-reviewed papers
incl. 1 Best Paper Award
0K+
HMDA records audited
algorithmic-bias forensics
+0.0%
Medical-IoT uptime
safety-aware energy budgeting
0%
Fairness-aware accuracy
mortgage-approval model
Portfolio tracker with AI-driven analysis
Personal portfolio tracker built with Next.js and Redis persistence. Includes an Oracle AI tab (Anthropic-powered analysis), position management, and a briefing generator that produces market context for active holdings.
Oracle AI analysis tab · Position management · AI briefing generator · Redis persistence · Real-time market context
Next.js · TypeScript · Redis · Anthropic API · Vercel · Real-time data
Personal fitness tracker with AI coaching
Daily fitness + nutrition tracker built with Next.js and Upstash Redis. Tracks workouts, macros, body metrics, and adherence to a 38-day sprint plan. Anthropic-powered weekly review and adaptive plan adjustments.
Daily workout + macro log · 38-day sprint adherence · Anthropic weekly review · Adaptive plan adjustments · PWA / offline-friendly
Next.js · TypeScript · Upstash Redis · Anthropic API · Vercel · PWA
Responsible AI · Fintech
Bias analysis pipeline on 800k+ HMDA records. Detected persistent demographic disparity in approval patterns; built fairness metrics dashboard and a fairness-aware classifier achieving 99% accuracy with documented demographic delta tracking.
Python · XGBoost · scikit-learn · Neural Networks · Fairness Metrics
Peer-reviewed work at the intersection of AI, healthcare, and safety-critical systems
ICICC 2026 · Springer Proceedings·Accepted·Best Paper Award·2026
An RL agent learns to dispatch renewable energy to surgical ICUs, balancing supply variability with hard uptime requirements on life-critical equipment.
Random Forest · XGBoost · LSTM · GRU · Reinforcement Learning· +2 more
AIMLA 2026 · IEEE (Machine Learning Track)·Published·2026
A Digital Twin layer allocates battery power across sensing, edge inference, and transmission via a Clinical Urgency Score, with physics + ML predicting battery state of health.
Digital Twin Modeling · Safety-Aware Energy Budgeting (SAEB) · Physics-based Battery Modeling · ML-based SoH and RUL Prediction · Clinical Urgency Score· +2 more
AIES 2026 / ICMLT 2026·Under Review·2026
Sole-authored research on calibrated abstention in human activity recognition for high-stakes deployments.
Decision Theory · Uncertainty Quantification · HAR · Calibration · Abstention Learning
[10] / The close
Now at CSH, graduating December 2026 and open to full-time AI/ML roles starting then. Also open to research collaborations and fellowship opportunities. Reach out, I respond quickly.