AI/ML Project Portfolio

This portfolio website showcases my projects and technical expertise in cutting-edge technologies, including:

• Generative AI Modeling
• Machine Learning and MLOps workflows
• Retrieval-Augmented Generation (RAG)
• Large Language Models (LLMs)
• Other Deep Learning applications

Generative AI Modelling

- - - - - - - - - - - - - - - - - - - - - - - - -

Generative Adversarial Networks (GANs Framework) for Biomaterial Discovery
[Description: Developed a GAN-driven Object-Oriented Framework for scalable GAN integration.] Link

DiffuSphere - An Object-Oriented Diffusion Models Framework for Image Generation
[Description: A comprehensive framework implementing DDPM, CFG, CFG ++, EMA and Power Law Delay EMA techniques for versatile diffusion model applications.] Link

Biomaterial Discovery using Deniosing Diffusion Probabilistic Models (DDPM - Diffusion Models)
[Description: Developed an Object-Oriented DDPM-based pipeline for a scalable code environment.] Link

Variational Auto Encoders for Biomaterial Discovery
[Description: Implemented an Object-Oriented VAE-based pipeline for biomaterial discovery with standardized integration.] Link

Machine Learning & MLOps Workloads

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

Object-Oriented Framework for Machine Learning (Industry Standard & Best Practices)
[Description: Created an Object-Oriented ML Framework for modularity, reusability, and maintainability.(View any logs in the Artifacts folder for better understanding).] Link

Performance Evaluation and Predictive Maintenance of Semiconductor Manufacturing Machines
[Description: This project focuses on performance evaluation and predictive maintenance of semiconductor manufacturing machines using over 600 sensor features.] Link

Scalable Machine Learning Pipeline including Model-Based and SFS Feature Selection
[Description: Implemented a scalable ML pipeline with feature selection techniques.] Link

H1N1 & Seasonal Flu Vaccination Prediction
[Description: Predicts individuals’ likelihood of receiving H1N1 and seasonal flu vaccines using health data with machine learning models and ranked 20 out of 6500+ competitors.] Link

Implementing Industry Standard Methodologies for Cirrhosis Prediction
[Description: Developed a pipeline adhering to industry standards for cirrhosis prediction.] Link

Analysis & Prediction of Test Results from Patients Healthcare Data
[Description: Created an ML workflow for analyzing and predicting healthcare test results.] Link

End-to-End MLOps implementation for Insurance Claims (Azure IaaC)
[Description: Created an end-to-end MLOps pipeline for automating the deployment and monitoring of machine learning models for insurance claims prediction.] Link

30+ Capstone Projects of DataScience and AI/ML
[Description: A collection of over 30 capstone projects covering various AI/ML topics] Link

Retrieval Augmented Generation (RAG)

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

RAG - Step-Back & HyDe (Semantic-Search) Pipeline for Complex Retreival
[Description: Combines semantic search with Step Back and HyDe techniques for more accurate retrieval in complex queries within RAG systems.] Link

RAG Query Translation Pipeline using Open AI & LangChain
[Description: This repository contains code for performing Query Translation techniques in RAG (Retrieval-Augmented Generation).] Link

RAG - Vector Search Pipeline for efficient retrieval
[Description: Utilizes vector embeddings for faster, similarity-based searches in RAG systems, enhancing retrieval efficiency.] Link

Other Deep Learning Workloads

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

Predicting the Attachment Level using Convolutional Neural Networks (Industry Standard Pipeline with Best Practices)
[Description: Implements an object-oriented pipeline for predicting attachment levels with CNNs, ensuring modularity, reusability, and scalability while adhering to industry best practices.] Link

Classifier Free Guidance for DDPM
[Description: A repository featuring Conditional Guidance Framework (CFG) and advanced diffusion models for generative modelling] Link