Zahra Abrishami
Los Angeles, California, United States
Summary
🤩
RockstarZahra Abrishami is a software and machine learning engineer with four years of experience building production-focused ML solutions and data pipelines, including a software engineering role on Meta’s Facebook Ads team. She holds an MS in Computer Science from USC (4.0) and has practical research-to-production experience from a Qualcomm AI Research internship, where she implemented batch normalization folding for the AIMET model-efficiency library to speed inference. Zahra has taught and led ML coursework at USC, designing assignments and managing automated grading and course infrastructure, which reflects strong communication and mentoring skills. She combines hands-on optimization work on open-source tooling with applied systems engineering—automating ad-engagement metrics pipelines at scale—making her well-suited for roles that bridge ML research and production engineering.
4 years of coding experience
Github Skills (13)
net10
quantization10
keras10
quants10
machine-learning10
lossless-compression10
deep-learning10
tensorflow10
python10
compress10
model-optimization9
compression7
open-source5
Programming languages (1)
Github contributions (2)
quic/aimet
Jul 2021 - Aug 2021
AIMET is a library that provides advanced quantization and compression techniques for trained neural network models.
Role in this project:
ML Engineer
ML EngineerContributions:10 commits, 13 PRs, 6 pushes in 1 month
Contributions summary:Zahra's contributions primarily focus on enhancing the AIMET library, specifically targeting batch normalization folding within the TensorFlow Keras framework. They implemented functionalities to identify patterns suitable for batch norm folding, including support for sequential, functional, and combined model architectures. The user also developed utilities to accurately identify and handle conv-bn pairs for optimization. Their work contributes to model compression and optimization techniques within the context of deep learning.
pytorchtechniquesdeep-learningpruningcompression
quic-zabrisha/aimet
Jul 2021 - Aug 2021
AIMET is a library that provides advanced quantization and compression techniques for trained neural network models.
Contributions:14 branches in 1 month
pytorchtechniquesdeep-learningpruningcompression
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.