Building scalable backend systems and ML-powered applications with a focus on performance, distributed architecture, and production-ready deployments.

Production-grade backend inspired by YouTube, designed with scalability, caching, and distributed processing patterns.

• Scalable Architecture: Engineered 40+ REST endpoints with Nginx load balancing (least_conn) distributing traffic across multiple Node.js instances.
• High-Performance Caching: Implemented Redis cache-aside strategy and a buffered view-count pipeline synced via cron jobs to reduce database write load.
• Database Optimization: Built complex MongoDB aggregation pipelines and implemented a persistent counter pattern reducing subscriber lookup from O(N) → O(1).
• Secure Authentication & Media: JWT authentication with token rotation and scalable media uploads using Multer + Cloudinary.
• Containerized Deployment: Dockerized services enabling consistent environments and easy horizontal scaling.

Tech Stack

Node.js • Express.js • MongoDB • Redis • Nginx • Docker • Cloudinary

Scalable backend system supporting multi-vendor ordering, geospatial restaurant discovery, and production deployment.

• Multi-Tenant Platform: Vendor onboarding, menu management, cart operations, and distributed order processing.
• Geospatial Search: Implemented GeoDjango + PostGIS queries for radius-based restaurant discovery.
• Production Infrastructure: Containerized with Docker + Nginx + Gunicorn, deployed on GCP VM with CI/CD via GitHub Actions.
• Payment Integration: End-to-end order lifecycle with Razorpay payment processing.

Tech Stack

Django • PostgreSQL/PostGIS • Docker • Nginx • Gunicorn • GCP • GitHub Actions

Production-ready backend for a social platform.

• JWT authentication with modular API architecture
• Database migrations with Alembic
• >95% endpoint test coverage with Pytest

Tech Stack

FastAPI • PostgreSQL • SQLAlchemy • Pytest • Docker

End-to-end ML system from training → serving → frontend deployment.

• Custom PyTorch UNet for automatic polygon coloring
• FastAPI inference API deployed on AWS EC2
• Netlify frontend with drag-and-drop UI

Tech Stack

PyTorch • FastAPI • Docker • AWS EC2 • JavaScript

Custom conditional UNet architecture with learned color embeddings.

• Synthetic dataset generator
• End-to-end training pipeline
• Vision model experimentation

Tech Stack

PyTorch • OpenCV • Computer Vision

Educational implementation of neural networks from scratch.

• Built an automatic differentiation engine
• Implemented MLP achieving 95% accuracy on the Breast Cancer dataset

Tech Stack

Python • NumPy • Automatic Differentiation • Neural Networks

Implementation of advanced Retrieval-Augmented Generation pipelines.

• Multi-query RAG, RAG Fusion, HyDE, RAPTOR, CRAG
• Query routing and relevance grading with web fallback

Tech Stack

LangChain • OpenAI • Vector Databases

Django • FastAPI • Node.js • Express.js • REST APIs

PostgreSQL • PostGIS • MongoDB • MySQL • SQLite • Redis

PyTorch • TensorFlow • OpenCV • scikit-learn • LangChain • LangGraph

Python • JavaScript • C/C++ • SQL

Docker • GitHub Actions • AWS EC2 • Google Cloud Platform • Nginx

• LeetCode Knight — Top 4.99% globally