Projects
My work spans spatio-temporal machine learning, network science, and computational social systems. I build models that learn from data distributed across space, time, and network structures, and apply them to real-world systems such as cities, elections, mobility networks, and scientific collaboration. Selected projects below highlight research contributions, open-source implementations, and production-ready systems built for scientific and policy-driven applications.
Research Projects
BrainLab OS: Learning from Minimal Data under Distribution Shift
Master's thesis developing theoretical and practical frameworks for robust machine learning under distribution shift and limited supervision. Building neuroanatomically-inspired cognitive architecture that learns efficiently from minimal labeled data across multiple domains.
Key Highlights
- •Developing theoretical foundations for generalization under distribution shift
- •Implementing neuroscience-inspired learning mechanisms for data-efficient training
- •Investigating robustness and adaptation in low-data regimes
Technologies
Neural Time Capsule: Urban Growth Prediction
ConvLSTM-based deep learning system for multi-decadal urban expansion forecasting using satellite imagery. Dual-channel architecture processes built-up surface density and road network data to predict city growth patterns 25+ years into the future.
Key Highlights
- •Achieves 0.000218 MSE with 67% improvement over U-Net baseline
- •Trained on 2,313 tiles spanning 1975-2000 across diverse U.S. regions
- •Lightweight 470K parameter model trains in 6 hours on CPU
Technologies
Disaster Risk Monitoring Using Satellite Imagery
End-to-end machine learning system for disaster risk assessment using satellite imagery. Jupyter notebook pipeline covering data preprocessing, efficient model training, and deployment for inference with UNOSAT flood event case study.
Key Highlights
- •Automated satellite imagery preprocessing pipeline
- •Efficient model training for large-scale disaster monitoring
- •Real-world UNOSAT flood event case study validation
Technologies
Bihar 2025: Real-Time Election Forecasting
Production-grade election forecasting system combining Bayesian hierarchical models, ensemble learners, and transformers for constituency-level predictions. Ingests live polls, social media sentiment, and news feeds with 5-minute update latency using streaming architecture.
Key Highlights
- •Multi-task ensemble for seat-level win probability and vote margins
- •Transformer-based NLP for real-time social media sentiment analysis
- •Streaming data pipeline with sub-5 minute prediction updates
Technologies
Selected Engineering Projects
COVID-19 X-Ray Classification with HPC
High-performance computing implementation of CNN-based COVID-19 detection from chest X-rays. Mac-optimized training pipeline leveraging Accelerate framework for efficient model training on Apple Silicon.
Key Highlights
- •CNN architecture for COVID-19 detection from chest X-rays
- •HPC-optimized training pipeline for Mac M-series chips
- •Automated preprocessing and data augmentation
Technologies
CLAD-PV: Cyber-Physical Security for Solar Systems
Physics-guided intrusion detection system for solar inverter networks using SunSpec/Modbus protocols. Mac-native implementation combines power system physics knowledge with anomaly detection to identify cyberattacks on distributed solar infrastructure.
Key Highlights
- •Physics-guided IDS for solar inverter security
- •SunSpec/Modbus protocol implementation
- •Detects power injection attacks and data manipulation
Technologies
CPS Threat Modeling for Solar Inverters
Comprehensive threat modeling framework for grid-tied solar inverter networks. STRIDE/DREAD risk analysis with architecture diagrams, gap analysis, and security recommendations for cyber-physical solar systems.
Key Highlights
- •Complete architecture diagram for solar inverter networks
- •STRIDE/DREAD threat matrix and risk assessment
- •Gap analysis with security recommendations
Technologies
Global Product Catalog: Multi-Region Distributed System
Production-grade distributed product catalog using DynamoDB Global Tables for multi-region replication. Terraform-managed infrastructure with comprehensive load testing demonstrating global consistency and sub-50ms latencies.
Key Highlights
- •DynamoDB Global Tables for multi-region active-active replication
- •Infrastructure-as-Code with Terraform
- •k6 load testing with 1000+ concurrent users
Technologies
Customer Churn Prediction System
Machine learning system for predicting customer churn in subscription services. Dockerized deployment with comprehensive data preprocessing and multiple model architectures for production-ready churn forecasting.
Key Highlights
- •Ensemble ML models for churn prediction
- •Dockerized deployment pipeline
- •Feature engineering for subscription data
Technologies
Aether: AI Therapist Web Application
Emotionally-aware AI therapist web app using Hugging Face Transformers for mental health support. FastAPI backend with transformer models provides empathetic, context-aware conversations for emotional well-being.
Key Highlights
- •Transformer-based conversational AI for mental health
- •Emotionally-aware response generation
- •FastAPI backend with web frontend
Technologies
Tripsy: AI-Powered Travel Planning
Intelligent travel planning system with backend API, frontend interface, and infrastructure automation. Features AI-powered itinerary generation, geospatial mapping, and user management.
Key Highlights
- •AI-powered travel itinerary generation
- •Interactive maps with Leaflet integration
- •Multi-component architecture (backend, frontend, infra)
Technologies
Global Airport Dashboard
Interactive web-based visualization dashboard for global airport data. D3.js-powered interface with geospatial mapping, statistical analysis, and filtering capabilities for exploring worldwide airport infrastructure.
Key Highlights
- •Interactive geospatial visualization of global airports
- •D3.js-based charts and statistical analysis
- •Real-time filtering and search capabilities
Technologies
NASA Logs Billing Dashboard
Credit-based billing system that processes NASA HTTP logs as billable events with real-time usage visualization. Dockerized Flask application with Plotly dashboards for usage analytics and invoice generation.
Key Highlights
- •Real-time event ingestion and billing processing
- •Interactive usage dashboard with Plotly visualizations
- •Credit-based billing with date-range queries
Technologies
LocalMapReduce: Distributed Computing Framework
Local MapReduce simulation using Python and LevelDB for distributed data processing. Implements map-reduce paradigm with persistent storage using Plyvel for efficient key-value operations.
Key Highlights
- •MapReduce implementation with Python
- •LevelDB for persistent key-value storage
- •Modular architecture with map and reduce phases
Technologies
Haven: AI-Powered Personal Finance Assistant
Privacy-first AI personal finance platform with expense categorization, budget recommendations, and financial insights. Local-first architecture ensures user data never leaves their device while providing intelligent financial guidance.
Key Highlights
- •ML-powered expense categorization with 90%+ accuracy
- •Personalized budget recommendations using spending patterns
- •Local-first architecture for maximum privacy