Hello, I'm Ali Shobeiri 👋
Thanks for visiting, here you'll find some information about me, my work and how to get in touch
About
How did I end up here...?
I am originally from Tehran, Iran. I grew up in Calgary, Alberta, Canada and attended McGill University to study Electrical Engineering. During my time at McGill I had the opportunity to work with a lot of great people and was involved in several organizations. My proudest achievements are leading McGill CodeJam, co-founding Hack4Impact McGill and teaching as well as organizing (with a lot of help) a peer taught machine learning course for 60 undergraduate students as part of the McGill AI society.
During my time at school, I also had great opportunities to travel, attending conferences, case competitions and hackathons. One of the most memorable trips for me is having a chance to compete and place second in a global business competition in Shanghai right before COVID-19 pandemic took off.
I am really interested in building cool technical projects and love to read books. If you have any book recommendations, or I can help with anything please don't hesitate to reach out. :)
Where I've worked
Machine Learning Engineer @ Apple
July - Present
- Working as part of the proactive intelligence team delivering machine learning directly to iOS devices.
Software Engineer @ Microsoft
September - July 2021
- Worked as part of the Windows Admin Center team.
- Launched a GPU extension for the platform that allowed users to manage and assign their GPU resources to virtual machines.
- Developed a platform autoreload functionality that significantly reduced development time by automatically reloading the platform after any development changes as opposed to the manual reloads required before.
- Added automatic documentation tool that would automatically document all SDK functionality for external and internal users.
Research Software Engineer in Machine Learning Intern @ Unity Technologies
May - Aug 2020
- Researched and developed machine learning aided animation tools to significantly speed up 3D character animation.
- Web scraped over 1500+ unique animation clips, increasing dataset sample size by over 50%.
- Implemented robust run-time data augmentation including mirrorin and random rotation strategies leading to 15% improvement in model loss performance.
- Developed an efficient forward kinematics loss enabling runtime conversion between joint rotations to joint positions thus allowing for direct optimization of positional outputs.
Machine Learning Engineer Intern @ Ubisoft
May 2019 - Aug 2019
- Researched a multi-agent navigation algorithm, along with obstacle avoidance using a multi-agent approach to Soft Actor-Critic.
- Implemented radar based agent perception leading to a 120% improvement on previous model's obstacle avoidance.
- Introduced new reward function leading to 40% improvement in agent path straightness and reduction in training time.
- Recreated game engine in Python, including randomized obstacles and navigation mesh generation leading to 10x reduction in model training speed and allowed for faster prototyping.
Data Science Intern @ Behaviour Interactive
January - April 2019
- Preprocessed a year of in-game data to train a machine learning skill estimation algorithm which showed 75% accuracy in predicting game outcomes.
- Developed internal web scraping tool that aggregated data about the video game industry and used Random Forest classification to predict the success of future games with 67% F1 score on a four class classification dataset.
- Aggregated months of historical data to determine optimal dedicated server locations to reduce latency for over 50k daily players.
Software Engineer Intern @ SAP
May - Aug 2018
- Prototyped search functionality for one the largest bank in Canada using Flask (Python) to service requests and serve most relevant queries using natural language processing.
- One of two developers working on developing a web interface using SAPUI5 for resource planning software used in a continental infrastructure project.
Software Engineer Intern @ Plotly Technologies
May - Aug 2017
- Was member of the team for the launch of Dash, a product which grew company revenues by more than 50%.
- Expanded Dash's component library, adding the DatePicker components. Also wrote a tool that would automatically generate documentation for all existing components and their properties, improving their future maintainability.
- Developed one of the first customer examples of Dash, which interactively filters and visualizes all New York City Uber ride locations for 2014, over 1GB of data.
- Received letter of recommendation from company CEO.
Projects
