Karanjot Singh Saggu
Ai Security Analytics Developer @TrendAI
Signup · Get unlimited contacts
WORK HISTORY
Ai Security Analytics Developer @TrendAI
Ottawa, ON, CA
EDUCATION
Punjabi University
Bachelor of Technology - BTech, Computer Science
University of Ottawa
Masters of Computer Science Concentration in Applied Artificial Intelligence, Computer Science
ABOUT KARANJOT SINGH SAGGU
Currently a full-time student in Masters of Computer Science Concentration in Applied Artificial Intelligence at the University Of Ottawa. Some of the work I have done is as follows:Professional Work-1. Implemented a 16-class document classifier (80% accuracy) and unified OCR (85% accuracy) for documentclassification and transcription of both printed and handwritten text for a global client. Deployed theimplementation using Docker, Flask, and Amazon Web Services.2. Designed and implemented an automatic text extraction from site inspection reports (pdf) using SentenceTransformer and tesseract OCR with an average 0.91 Jaccard score per document for an end US-based client speedingup the process by 7 times (40 minutes per document to just 2 minutes per document). Deployed the model using Docker, Fast-API, and Amazon Web Services.3. Designed and implemented a house-inspection matcher to predict work items required for an inspection job usingnovel in-house Deep Learning Architecture improving the performance from 31% to 85% for a US client. Deployedthe model using Docker, Fast-API, and Amazon Web Services.5. Extensive work on object detection models like Faster-RCNN, YOLO, etc.Self-learning and University Projects-1. Implemented a mini version of a computation graph identical to PyTorch for training a neural network.2. Worked on Variational Auto-encoders, Generative Adversarial Networks, and Wasserstein GANS to understand generative modeling.3. Worked on Adversarial Training to train models that are invariant to random perturbations in the image thus improving the prediction accuracy.4. Implemented and Deployed a detector model that detects the age and ethnicity of the person using a multi-task Convolutional Neural Network using Fast-API and Amazon Web Services EC2 instance.International Publications-1. Knowledge Graphs for Recommender System:(Page 212-217) FCDEGS International Conference.2. 3D-Multi Object Detection – A Review (Page 260-266) FCDEGS International Conference
This profile is compiled from publicly available professional sources. Unifers is not affiliated with or endorsed by LinkedIn. Request removal of this profile.