San Jose State University

I graduated with a Master of Science in Artificial Intelligence in May 2023. As a primary subject in my graduate studies, I focused on Intelligent Autonomous Systems, AI & Data Engineering, Machine Learning, Data Mining, Natural Language Processing (NLP), and Deep Learning.

Gujarat Technical University

As part of my pursuit of a Bachelor of Engineering in Computer Science and Engineering, I focused on learning artificial intelligence, Python, data structures and algorithms, Java, database management systems, and C++.

AI Graduate Research Assistant

@ SJSU Research Foundation

- Created a small autonomous vehicle by modifying the F1/10 car to provide mounting points for components like 2D Lidar, Teensy board, IMU 6500, and ZED stereo camera to behave 100% autonomously leveraging the Nvidia TX2 board.
- Developed custom neural network models using TensorFlow to improve object detection accuracy by 15% over pre-trained models.
- Integrated the ZED Stereo Camera with ROS and Docker, reducing system latency by 20% and improving overall performance.
- Implemented a custom algorithm to process sensor data from 2D Lidar, Teensy board, IMU 6500, and ZED stereo camera with a 98% accuracy rate in obstacle detection and avoidance.
- Developed and implemented a novel algorithm that improved vehicle localization precision by 20% using 3D point clouds from a visual odometry stack, leading to a more accurate mapping in real-time.
- Computed 12 different state-of-the-art models for Traffic Signs dataset and optimized the speed of these models’ training by 12% using Intel’s integration for computation. Optimized the speed of inference by 10 times with INT8 scales of TensorRT.
- Created Django based webpage for live streaming of a robot’s video feed with reduced latency time by 6 milliseconds.
- Managed to run client-side JavaScript to capture user input for different features, received it on the server side processed it to controlled motors, sensors, and various functionality over the internet using WebSockets.

Software Engineer (AI/ML Department)

@ Startlit Electronics

- Collaborated with cross-functional teams to design and execute a successful proof-of-concept (POC) for the home security product, saving 10% on the total POC budget while delivering all required functionalities within the desired timeline.
- Developed and integrated machine learning algorithms to enable motion following and weapon detection features in a home security product prototype, resulting in a 15% increase in accuracy compared to existing products on the market.
- Innovated custom neural network architecture with a transfer learning approach, achieving a precision rate of 93% for people detection.
- Led development for Natural Language Processing (NLP) initiative with chatbots and virtual assistants for home security applications that worked on language understanding methods like a bag-of-words and TF-IDF.
Used the AWS SageMaker and S3 to quickly build, train and deploy the models like YOLO, Regression, SVMs, and Decision Trees.
- Performed data exploratory analysis, data visualizations, and feature selections using Python and Pandas.
- Developed a predictive model for product sales using logistic regression which accurately forecasted sales within 2% of actual figures, leading to increased profitability and better inventory management.
- Worked closely with and extended my support to the Tech Lead and the Project Manager to hire, train, and onboard two new team members.