Levent Özbek
Sr. Machine Learning Engineer & Architect.
Applied Quantum Scientist.
Optimization Expert.
First person to ever bring AI to Visual Effects.
A Sr. Machine Learning Engineer with experience in architecting and deploying ML systems across various industries, including Quantum Computing, Visual Effects and Digital Advertisement. Adept at building and productionizing models, such as Diffusion, GANs, LLMs, and RL. A pioneer in bringing AI to VFX, introducing generative AI to enhance CGI asset creation and screenwriting. Demonstrated expertise in Quantum ML, contributing to Canada’s first team to train and test an LLM on a Quantum computer. Strong background in Deep Learning, optimization, and HPC, with experience deploying models for image generation, inpainting, and outpainting. Proven ability to leverage photonics for optimization and developed battery simulations.
Open Source Projects
Projects will be dynamically loaded here...
Education
University of London, Goldsmith
Hons. BSc. in Computer Science
2019-2023 | GPA: 3.7/4.0
University of Leeds
MSc. in Artificial Intelligence
2024-2025
Industry
Machine Learning Engineer
Alberta Machine Intelligence Institute - Edmonton, AB (Remote) | April 2024 - Present
- Built Torchlaunch, which is a Pytorch wrapper that makes distributed training easier for researchers.
- Was part of the founding team behind Amii’s Vulcan clusters, as part of the PAICE initiative.
- Wrote the first Slurm config file for the Vulcan clusters.
- Worked on implementing a smart resource allocation system for Slurm job submissions.
- Created ML training materials for researchers and residents.
Founding Machine Learning Engineer (Diffusion Models)
Adora - Seattle, WA (Remote) | April 2024 - January 2025
- Built & deployed a custom inpainting model in the context of advertisement creatives.
- Built & deployed a custom outpainting model to mimic Adobe’s “infinite canvas” feature.
- Built & deployed an image generation pipeline for travel advertisements.
- Built & deployed a RHLF (Reinforcement Learning from Human Feedback) router to route prompts to appropriate models, LoRA, and config parameter settings.
Lead Research Engineer (Generative AI)
Crafty Apes VFX - Montreal, QC (Remote) | November 2022 - April 2024
- First person to bring generative models to the visual effects industry.
- Built & deployed a diffusion-based CGI asset generator.
- Built & deployed an LLM that acts as a screenwriting assistant tool.
- Built & deployed a prompt-based face swap tool.
Quantum Research Resident (Quantum Optimization)
Creed & Bear - Remote | November 2023 - April 2024
- Researched leveraging Quantum computing & photonics for optimization of energy distribution.
- Built a nonreciprocal quantum battery simulator using Qiskit & Pennylane.
- Developed a Monte Carlo Quantum circuit.
Machine Learning Quantitative Research Scientist
Borsa Istanbul - Remote | June 2022 - October 2022
- Boosted alpha generation and portfolio sharpe ratio
- Reduced model training time by 30%
- Improved execution efficiency
- Identified and monetized latent market opportunities
- Achieved 80% improvement in hit rate
- Increased production system reliability
Lead Applied Scientist
Netflix - Vancouver, BC | March 2021 - May 2022
- Lead the ML team in this newly acquired Visual Effects studio, working with the data acquisition team.
- Built & deployed a GAN to remove motion blur from 1.5 million images.
- Built & deployed an instance segmentation UNet model to parse human body parts & clothing.
- Adapted a NeRF model to create short videos from a limited number of image stills.
Quantum Machine Learning Research Intern
IBM - Ottawa, ON (Remote) | June 2021 - September 2021
- Collaborated with the quantum team to optimize LLM training and inference on a quantum computer.
- Built a single-head attention mechanism using a series of quantum circuits.
- Tech stack included Python, Qiskit, and Pennylane.
Machine Learning Engineer Intern
Stream Hatchet - Barcelona, Spain | December 2020 - March 2021
- Built a synthetic dataset of logos in livestreams using the Twitch API.
- Built a custom augmentation pipeline for logos placed on screenshots.
- Trained a YOLOv4 on this synthetic dataset, achieving a mean average precision of 97.2%.
Data Engineer
VESTEL - Montreal, QC (Remote) | May 2017 - October 2020
- Set up and maintained the data infrastructure for Vestel and subsidiaries across Turkey and Europe.
- Worked heavily with MySQL and Hadoop integration.
Software Developer Intern
VESTEL - Montreal, QC (Remote) | February 2016 - February 2017
- Worked on SQL databases and automated query scripts in SQL.
- Submitted weekly reports.
Awards
Renewable Bachelor Degree Scholarship
Year: 2017
FIRST Robotics Waterloo Regional Winner (Gold Medal)
Year: 2009