Ali Jahani Email & Phone Number

Toronto, ON, CA

San Francisco, CA, US

- Working on inference and improving GPU performance for genAI

Agentic RAG platform to speed up legal research for lawyers using AI!

Kitchener, Ontario, CA

• Focusing on MLOPs and ML infrastructure to facilitate the ML projects lifecycle at Avidbots e.g. smart data collection, monitoring, training, experiment tracking, and model deployment
• Built a data-lake for our fleet to be used for future machine learning projects
• Designed and implemented Personal Identifiable Information Removal package for GDPR compliancy Technologies: Dagster, ROS, Tensorflow, Pytorch, Docker, Openvino, Apache Spark, Terraform, Kubernetes, Bitbucket.

Kitchener, Ontario, CA

• Semantic Segmentation Tech Lead - Deployed Avidbots’ first ML model in production
• Implemented a pipeline of data collection from robot to deployment
• Performed various pruning and quantization techniques to improve model inference speed by up to 90%
• Implemented a semi weakly supervised training to boost performance by 3%
• Designed and implemented an auto annotation tool using U^2 Net
• Implemented active learning-based image selection for annotationTechnologies: Tensorflow, Pytorch, Docker, Openvino, Streamlit, ROS, Flatland Simulation, Nvidia ISAAC, Python, C++, OpenCV, Pandas, MongoDB, CVAT

• (Acquired by Raven Industries, subsequently by CNH Industrial)
• Developed a 3D farm Simulator for testing and synthetic dataset collection
• Collected dataset in fields for combine object detection
• Fine-tuned state-of-the-art semantic segmentation models for a semi-auto annotation tool
• Technologies: Python, Tensorflow, Pytorch, Detectron2, OpenCV, ROS, Carla, Unreal Engine 4

Edmonton, Ab, CA

• Used LiDAR (as supervised) and Stereo images (as unsupervised) to improve state-of-the-art single image depth estimation accuracy by ~3%
• Proposed new evaluation based on cleaned ground truth LiDAR
• Integrated deep learning depth estimation with SLAM to recover scale and improve accuracy and robustness
• Taught Introduction to Computing Science to non-computer science students
• Technologies: Python, Tensorflow, Pytorch, OpenCV, ROS

Edmonton, Alberta, CA

• Implemented a rule-based AI and in-game hint system in multiplayer virtual reality escape room games
• Pitched and implemented new successful ideas, e.g. destructible meshes
• Performed various profiling and optimizations to reach 90 fps
• Technologies: Git, HTC VIVE, Unreal Engine 4

Master Of Science - Ms, Computer Science

B.Sc. In Electrical Engineering, Control Systems