Projects

Explore a collection of my software projects, ranging from cloud computing and web applications to data science and machine learning. Each project highlights my experience in designing, building, and deploying solutions that combine strong engineering practices with modern technologies.

Adaptive Game Agent for Ashta-Chamma Reinforcement Learning • Python • OpenAI Gym

Built an intelligent agent for Ashta-Chamma, a traditional Indian board game, using reinforcement learning algorithms. The agent adapts to different opponent strategies through Proximal Policy Optimization (PPO) and Deep Q-Networks (DQN) implementations.

Developed a custom training environment using OpenAI Gym that simulates various opponent behaviors including aggressive, defensive, and stochastic strategies. The system enables dynamic decision-making in high-dimensional state spaces by optimizing policy gradients and value functions for strategic gameplay.

Github →

CIFAR-10 Image Classification System Deep Learning • PyTorch • Neural Networks

Implemented multiple neural network architectures from scratch using PyTorch, achieving over 70% accuracy on the CIFAR-10 dataset. Built core deep learning components including vectorized backpropagation, batch normalization, layer normalization, and dropout regularization.

Conducted systematic experiments comparing various optimizers (SGD, Adam, RMSprop) and regularization techniques. Analyzed training dynamics and convergence patterns to optimize model performance across different CNN architectures.

Github →

Chess Match Database System Deep Learning • PyTorch • Neural Networks

Designed and implemented a key-value database system for storing and analyzing chess match data. Built tools to intake chess match movesets and player information, storing structured data to enable statistical analysis of match history, game openings, player matchups, and move sequences.

Developed query interfaces for retrieving statistics based on various criteria including opening variations, player performance metrics, and historical match patterns. The system heavily prioritizes reads, over writes.

Embedded Face Detection System Computer Vision • C++ • AdaBoost • Raspberry Pi

Led development of a real-time face detection system using the Viola-Jones algorithm and AdaBoost machine learning. Achieved 95% accuracy across 4,000 sample images while maintaining performance on resource-constrained hardware.

Implemented the system in C++ and deployed on a virtual Raspberry Pi to demonstrate real-time processing capabilities. Created automated testing pipelines using Bash scripting to streamline development and validation processes.

Github →

Cloud Infrastructure & Monitoring Platform AWS • Docker • Kubernetes • Prometheus • SNS

Deployed and managed a multi-service web application across different cloud environments, starting with Docker Compose on EC2 and scaling to a production-ready EKS cluster. Configured the Kubernetes environment with multiple worker nodes and established proper networking, demonstrating end-to-end containerization and orchestration workflows using existing application manifests.

Implemented comprehensive monitoring and alerting infrastructure using Prometheus to track cluster health metrics and pod performance. Built an automated notification system integrated with AWS SNS that triggers email alerts when pod restart counts exceed predefined thresholds, ensuring proactive incident detection and response for production workloads.