Sri Nikhil
Bandi

ECE @ Penn State

About

I'm an Electrical and Computer Engineering student at Penn State with a passion for building things at the intersection of hardware and software. I enjoy working on embedded systems, robotics, computer vision, and full-stack projects. I'm always looking for opportunities to build things that matter.

Projects

Fortnite Reinforcement Learning Agent

Python · PyTorch · YOLOv8 · PPO · Gymnasium

Built a reinforcement learning agent that learns to autonomously eliminate guards using only real-time vision and hardware-level input control, without game memory or APIs. Trained an Actor Critic network with Proximal Policy Optimization on a minimal five-value observation space derived from a custom YOLOv8 detector, using a dense reward structure to shape detection, approach, aim, and shot accuracy.

FoodBridge

Python · XGBoost · Scikit-Learn · Pandas

Engineered an AI-driven "Intelligence Layer" to reduce food waste in university dining halls by predicting consumer demand. Developed a high-performance forecasting model using the XGBoost gradient-boosting algorithm, integrating historical dining data and external variables to optimize inventory procurement and minimize operational surplus.

Pathfinding RL Algorithm

Python · PyTorch · Reinforcement Learning

Developed a pathfinder where a deep Q-network learns to navigate a 2D grid and repeatedly reach a goal while avoiding walls. The agent observes local danger, direction, and goal position, and improves over thousands of episodes, visualized with Pygame and tracked with live training plots.

Reapo-AI

Python · Next.js · Supabase · GitHub API · LLMs

Hackathon project: Developed an "Intelligence Layer" for large-scale codebases using vector embeddings and semantic search. The system indexes entire GitHub organizations, allowing AI agents to perform cross-repository feature analysis, automated PR reviews, and complex refactors with context.

Experience

2025 – Present Research

Undergraduate Researcher

Penn State Laboratory for Perception, Action, and Cognition (LPAC)

Developing a multi-modal Transformer for high-fidelity keypose classification and human gait analysis. I focus on automating the synchronization of multi-modal datasets to streamline model training, while researching Active Vision Systems to bridge the gap between computer vision algorithms and robotic actuators in intelligent environments.

Contact