CERTIFICATIONS

Generative AI: Working with Large Language Models

AWS Machine Learning Speciality(P)

Introduction to Large Language Models (Google Cloud)

Generative AI Fundamentals (GCP)

AWS Certified Cloud Practioner(P)

Deep Learning Specialization

Natural Language Processing Specialization

ChatGPT Prompt Engineering for Developers

Courses
Natural Language Processing
Deep Learning
Machine Learning
Applied Statistics
Data Mining
Language models Capstone Project Big Data 
Data Analytics with R Programming
Data Visualization

Deployed, configured Llama2(Large Language Models) on AWS Sagemaker, and optimized it to ensure efficient utilization of resources and optimal performance of the model.

Diagnosing and Mitigating Bias in Large Language Models
• Evaluated biases in large language models (LLMs) such as BERT, GPT-2, etc., addressing gender, race, and cultural biases. [BLOG]

Speaker Classification using Transformers
• Applied Self-Attention mechanisms and optimized transformer parameters, on a speech dataset featuring 600 speakers, resulting in a 20% reduction in training time with a commendable categorization accuracy score of 96.85%, demonstrating expertise in advanced machine learning techniques.

Adversarial Attack
• Implemented non-targeted iFGSM from scratch in PyTorch on CIFAR-10 dataset to generate adversarial samples which successfully reduced the model's accuracy from 95% to 0%.

Mastering Lunar Landing: Cutting-edge AI with OpenAI's Gym
• Applied Policy Gradient algorithm using Deep Reinforcement learning to accurately guide the Lunar's landing.

Unveiling Culinary Delights with Image Classification Technology
• Trained a neural network model using a dataset comprising 9867 jpg images followed by hyperparameter tuning, optimizer selection, exploration of complex model architectures suitable for image classification, and consideration of cross-validation techniques to optimize model generalization

Revolutionizing Wireless Communication: User Localization Through Self-Supervised Learning in IoT Environments
• Explored the efficacy of self-supervised learning for user localization based on Channel State Information (CSI), aiming to reduce reliance on labeled data and accommodate changing radio environments.

• Utilized a dataset obtained through a massive MIMO channel sounder, comprising labeled samples with real and imaginary parts of estimated channel matrices, signal-to-noise ratios (SNR), and ground truth positions of transmitters, as well as unlabeled and test data.

Case Study: Quora Question pair similarity problem
Identified which questions asked on Quora are duplicates of questions that have already being asked.