Dev Mittal

A scalable platform to automate programming lab assessment from submission to grading with secure, isolated code execution at scale.

• ▸Architected a cloud-native evaluation engine handling concurrent submissions via Kubernetes Jobs on Minikube.
• ▸Built a Next.js + React frontend with SSR for real time code submission and result tracking.
• ▸FastAPI backend integrated with Supabase (PostgreSQL + Auth) for secure auth and persistent storage.
• ▸Dockerized execution environments to safely run untrusted student code with strict isolation.
• ▸Automated compilation, execution, and test case grading eliminating manual TA evaluation.

A full stack campus security portal managing student exit passes, family visits, employee guest passes, and walk in visitor registration with QR based verification.

• ▸Role-based access control with 6 roles using Supabase Row Level Security and server-side auth.
• ▸Multi-step approval workflow with automated email notifications via Resend.
• ▸Domain-restricted Google OAuth with automatic role assignment using PostgreSQL triggers.
• ▸Real-time QR code validation and entry/exit logging for gate security operations.

An end-to-end ML pipeline that extracts entity-level sentiment from financial news and forecasts stock returns across 1, 2, and 5-day horizons.

• ▸Curated and preprocessed a 25K+ entity-level financial news dataset for entity-aware sentiment extraction and temporal return forecasting.
• ▸Contributed to the end-to-end ML pipeline architecture, integrating the sentiment-extraction stage with downstream forecasting models.
• ▸Benchmarked multiple models — Ridge, XGBoost, LightGBM, and a PyTorch MLP — for forecasting performance comparison.