Mohammad Gharachorloo

Senior Machine Learning Engineer @SoundHound AI

Toronto, ON, CA

Full ownership of R&D projects from feasibility study to prototyping to production.• Maintaining and developing end-to-end pipeline solutions for training DL models for wakeup phrase detection.• Full development of torch2tflite conversion tools for reducing memory and CPU consump-tion.• Developing pretraining/finetuning framework for cost reduction and improved efficiency.• Developing augmentation pipelines based on customer needs and acoustic conditions.• Developing model profiling tools for monitoring CPU and Memory footprints.• Developing a variety of DL models such as CNN-RNN, fully connected and transformer-based.• Writing tests and sanity checks to ensure the accuracy and stability of the newly designed solutions.

K. N. Toosi University of Technology

Bachelor of Science (BS), Telecommunications Engineering

QUT (Queensland University of Technology)

Master of Engineering (MEng), Communication and network specialism