AI Engineer

github

srijith Chetla • Peerlist

Eco Bot is an AI-powered solution designed to tackle underwater plastic pollution. Using a YOLOv8 object detection model, it identifies and locates plastic waste in underwater images or live video feeds with high accuracy. What makes Eco Bot unique is its integration with GPS data, enabling real-time mapping of pollution hotspots. Detected locations are visualized on an interactive map using Leaflet.js, helping environmental researchers and cleanup teams prioritize areas with high plastic density

Eco-Bot (Underwater Plastic Detection using Y

<p>EventSphere is a full-stack event management platform. It allows users to create, manage, and register for events with secure authentication and role-based access control.</p>

EventSphere

This project implements a Neural Network from scratch in Python using only NumPy to recognize handwritten digits from the MNIST dataset. It’s an educational project designed to understand how neural networks work internally without relying on high-level libraries like TensorFlow or PyTorch.

Neural Networks from Scratch

• Analyzed a retail dataset (118k+ rows) and engineered features such as tax-adjusted price and seasonal time variables. • Applied Isolation Forest to detect anomalous price spikes/dips across years and product categories. • Trained regression and classification models; Random Forest achieved 83.13% price-tier accuracy. • Built a Power BI dashboard highlighting category trends, tax impact, and essential vs. non-essential price differences, with 15% faster insight discovery through interactive filters and drill-down visuals.

Retail Price Analysis & Prediction

<p>-Developed a multi-disease retinal classification system on 20,000+ fundus images detecting 6 eye diseases using an ensemble of Vision Transformer and binary CNNs, achieving 93.7% accuracy. -Applied knowledge distillation to compress the ensemble into a lightweight CNN, retaining 82.7% accuracy for efficient edge deployment. -Built end-to-end data pipelines with preprocessing and augmentation techniques to improve model robustness across diverse real-world retinal imaging conditions.</p>