I'm a Data Scientist who loves playing with data using smart thinking and tech skills. I've been doing this stuff for over 4 years now and am currently pursuing Masters in Computer Science at NYU (yes NYU!). I'm into tech like deep learning, machine learning and computer vision, using tools like TensorFlow, PyTorch, OpenCV and Explainable AI. I'm all about turning complicated data into useful insights in order to support data-driven decision making.
Quick Note: I make mistakes but quickly learn from it (think of me as an RL agent xD)
- Relevant Coursework: Design & Analysis of Algorithms, Machine Learning, Big Data, Deep Learning, Computer Vision, Foundation of Entrepreneurship (Stern)
- Relevant Coursework: Design & Analysis of Algorithms, SQL (Database Management Systems), Artificial Intelligence, Machine Learning, Operating Systems
-
Languages:
-
Frameworks/ Libraries:
-
Tools/ Technologies:
-
Operating Systems:
- Utilized Apriori algorithm for purchase analysis, discovering key product associations/patterns for upselling and cross-selling, potentially leading to a 15% increase in avg. revenue per customer.
- Implemented an end-to-end custom LLM system on top of OpenAI’s GPT-3.5 for answering questions based on purchase analysis. Used vector embeddings and Neo4j graph networks for faster queries.
- Developed an ensemble of Customer Lifetime Value (CLV) and churn prediction models, leading to data-driven customer segmentation & risk profiling with 94% accuracy.
- Designed a robust & scalable architecture to identify & classify 15 defect categories on stoppers using XceptionNet deep learning model, achieving an accuracy of ~92%.
- Deployed the defect detection deep learning model in production using Docker and Azure. Used MLFlow for version control and monitoring model performance.
- Created Power BI reports to visually represent & communicate daily classification results & insights to stakeholders, resulting in a 37% better identification of production line issues.
- Designed and set up ROS-Gazebo pipelines to train Turtlebot3 Waffle Pi for autonomous navigation in the manufacturing plant using DQN algorithm, achieving a 95% success rate.
- Reduced training time for Turtlebot3 by approximately 15% by integrating Human Intervention Learning, enabling the robot to learn directly from human experience stored in the buffer.
- Trained and deployed Particulate classification models for structured and unstructured data using Azure ML and Azure Function Apps. Achieved an overall accuracy of ~97%.
- Replaced the manual classification process used by the Lab Analysis team with the new models, resulting in improved accuracy and efficiency in particulate classification.
Feel free to reach out for collaborations or just a friendly chat:
🌐 While you're here, don't forget to check out my repositories below!