Projects
Some projects I have worked on.
RAG Core
A library designed to reduce the implementation effort for Retrieval-Augmented Generation applications to a configuration file and a four main methods. Contributions are welcome!
Learn moreDocInsights platform
A platform which allows you to upload documents, and search through your library using natural language. Uses retrieval-augmented generation, LLMs, embeddings, and is deployed in Google Cloud as micro-services.
Learn moreCOCO Traffic dataset
The Common Objects in Context (COCO) dataset is a popular choice to train object detectors. COCO Traffic refines and enhances the classes for traffic lights to distinguish their state.
Learn moreDriver Assistant iOS App
A real time object detection app to detect traffic signs for iOS. This end-to-end project begins with the curation of a dataset, COCO traffic, explains the model training, and deployment in an iOS app.
Learn moreStyle Transfer Web App
This end-to-end project showcases how to design and tune a deep learning computer vision model to deploy it in a browser using TensorFlow JS. Since the model must run on the client side, it is quite resource constraint.
Learn moreObject Detection Basics
History of object detection models, starting from feature extractors like HOG, and detailed explanations of convolutional neural networks, metrics to evaluate them, and the YOLO, Faster-RCNN, and DETR architectures.
Learn moreAdversarial Examples
How neural networks are susceptible to small, imperceptible changes called adversarial attacks, and what this means for safety and the learned representations. Includes implementations of common attacks.
Learn more