Amir Abdi
Principal Applied Scientist at Offerland
Vancouver, British Columbia, Canada
Summary
🤩
Rockstar🎓
Top SchoolAmir Abdi is a Principal Applied Scientist based in Vancouver with a PhD in Machine Learning from UBC and over a decade of hands-on experience turning research into production AI. He has led teams and productized time-series forecasting, self-supervised learning, and model selection/abstention at Borealis AI and now drives applied research at Microsoft while co-founding and CTOing a startup. His work spans from building BorealisAI’s first centralized feature store to practical model deployment tools—he maintains tooling to convert Keras models to optimized TensorFlow inference graphs, including quantization and serialized graph outputs. Comfortable teaching and mentoring, he also brings an unusual background as a trained dentist and long-term RoboCup technical committee contributor, reflecting a blend of rigor, interdisciplinary problem-solving, and systems-level thinking.
12 years of coding experience
3 years of employment as a software developer
Doctor of Philosophy (PhD) Elec & Computer Engineering - Machine Learning, Doctor of Philosophy (PhD) Elec & Computer Engineering - Machine Learning at The University of British ColumbiaBachelor's degree Computer Software Engineering, Bachelor's degree Computer Software Engineering at Shahid Beheshti UniversityDoctor of Dental Surgery, Doctor of Dental Surgery at Shahid Beheshti University of Medical SciencesMaster's degree Computer Engineering - Artificial Intelligence, Master's degree Computer Engineering - Artificial Intelligence at Sharif University of Technology
Stackoverflow
2,377reputation
261kreached
33answers
14questions
ssh
top-5%
pycharm
top-5%
Github Skills (17)
python10
machine-learning10
keras10
tensorflow10
model-conversion10
continuous-deployment9
ml-deployment9
graph9
visual-studio9
optimisation9
optimization9
ssh9
pycharm9
quantization9
database6
Programming languages (10)
Github contributions (5)
amir-abdi/keras_to_tensorflow
May 2017 - Nov 2020
General code to convert a trained keras model into an inference tensorflow model
Role in this project:
ML Engineer
ML EngineerContributions:45 commits, 10 PRs, 35 pushes in 3 years 7 months
Contributions summary:Amir's contributions centered around converting a Keras model to a TensorFlow frozen graph for inference. They developed a script that loads a Keras model, renames the output nodes, and converts variables to constants. The user also added functionality for quantizing the weights of the model and included features for saving the graph definition in ascii format. They also implemented a method for loading the model architecture through either json, or yaml formats.
keras-modeldeep-learninginferencemachine-learningtensorflow-inference
amir-abdi/artisynth_point2point
Jan 2018 - May 2019
Deep Reinforcement Learning for Biomechanical Modeling
Contributions:94 commits, 2 PRs, 19 pushes in 1 year 3 months
deep-learningreinforcement-learningdeep-reinforcement-learningmodelingreinforcement
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.