Featured Projects

A showcase of my best work and technical achievements

LapTrack iOS App
Source Code

LapTrack

iOS Application

  • Developed a fully functional iOS application, LapTrack, using Swift and Xcode, integrating CoreML for backend data processing.
  • Designed an intuitive UI with optimized performance for a seamless user experience.
  • Implemented an ML model to efficiently store and manage data, ensuring fast and accurate predictions.
  • Optimized the Model using Random-Forest, XGBoost Regressor and Column Transformers followed by Pipeline.
Swift Xcode CoreML ML Model
Source Code
Harvest.AI Platform

Harvest.AI

AI-Powered Agricultural Platform

  • Spearheaded the development of an AI-powered agricultural platform to tackle critical challenges such as extreme temperatures, water scarcity, and nutrient-deficient sandy soil, driving improved crop yield and resource efficiency.
  • Engineered a seamless integration of cutting-edge AI/ML models with an intuitive user interface, leveraging ChatGPT-4 Plugins via REST API for enhanced user interaction and intelligent decision-making.
  • Designed and deployed scalable AI-driven solutions for precision farming, optimizing irrigation, fertilization, and climate adaptation strategies to promote sustainable agriculture through data-driven insights.
AI/ML TypeScript Next.js Recoil RapidAPI LightGBM
Source Code View Live Documentation
Semantic Segmentation Project

Semantic Segmentation of Gastric Polyps

Medical Image Analysis

  • Developed an AI-driven diagnostic tool for the early detection and segmentation of gastric polyps in endoscopic images, aiding in the timely identification of potentially cancerous growths.
  • Created a comprehensive dataset by generating annotated files, bounding boxes, and pixel-wise segmentation masks, ensuring precise detection and classification of polyps.
  • Implemented a DeepLabV3+ model with a MobileNetV2 encoder, leveraging Dilated Spatial Pyramid Pooling (ASPP) to capture multi-scale contextual information for accurate segmentation.
  • Preprocessed endoscopic images through contrast enhancement, noise reduction, and normalization to improve model performance and enhance feature extraction.
  • Designed a lightweight yet powerful model, optimizing computational efficiency for real-time inference in medical imaging applications, enabling faster and more accurate diagnoses.
MobileNetV2 DeepLabV3+ PyTorch OpenCV Medical Imaging Deep Learning
Source Code