Experienced Developer with a demonstrated history of working in the AI industry. Have worked on Aerial Imagery, Construction Imagery, Retail, and Medical Imagery.
Area of proficiency: Python, Numpy, Pandas, Deep Learning, RL, Keras, DevOps, C++, AWS, TF, OpenVino, Unity-3D
Worked on retail analytics, handled 1000 cameras, deployed async AI models for analytics.
• Received Top performer (Future Star Band- 4.44/4.50) in the CV/ML Team.
• Built Dataset generation pipeline for object detection and semantic segmentation using Unity3D, which included domain randomization techniques
• Developed a simulated environment in Unity-3D for autonomous flight and trained the RL model. Improved the reward using different randomization techniques, gamma, and stacked observations
• Created a Python module for visual odometry using sensor fusion for path planning in GPS denied situations
• Core member of first CV-ML Team, which raised 1M$.
• Built communication module for transferring RTSP stream and meta-data from Raspberry-PI, Jetson- Nano to android smartphone.
• Built Keypoints prediction model using Keras for pose estimation of an object.
• Built model for Surface Material Classification using Deep Learning (Keras).
• Built a Machine learning model for Activity Prediction using IMU data.
• Built Highly Precise Automatic Number Plate Recognition and classification.
• Built Highly efficient multi Threaded GUI Application for Construction Site Insights(Truck count, PPE, Person Count, Truck Movement analysis using GPS HAT module for raspberryPi) on top of state-of-the-art Object Localization Techniques using OpenVino and NCS stick.
• Created python Classes for sequential data upload to AWS S3 and software update for edge device software.
• Created Data Reconciliation scripts for daily asset metrics(Incomplete Transactions, Missing packets, etc.).
•Automation of server provisioning and configuration using custom templates with Ansible.
•Trained Yolo v2, v3 model using custom images.
•Annotating of images for YOLOv2, YOLOv3, and data augmentation to increase dataset size.
•Object detection and classification of dataset crawled from various sources.
•Setup ELK stack for Application log analysis.
•Using MQTT broker and subscriber fetched images from raspberry pi on the AWS server and pushed classification results to S3.
•Using Signal Processing and Image processing Algorithms developing a tool to Convert Level 0 Raw SAR Data to Level 1 for human footprint classification
Grade: :D
Grade: 93%
Grade: 87%
Apart from being an AI developer, I enjoy meeting new people and building stuff. I am an aeromodeller and practice drone racing during my free time. During the warmer months here in Canada, I enjoy playing basketball and running.
I am an aspiring chef, and I spend a lot of my free time exploring the latest technology advancements in the AI development world.