Philipp Guevorguian Email & Phone Number

Yerevan, Armenia

• Led the development of scalable, distributed multi-billion parameter machine learning architectures leveraging PyTorch and advanced techniques such as Fully Sharded Data Parallel (FSDP), torch compile, hardware aware.
• Co-designed novel model-in-the loop methods for molecular design via discrete optimization achieving SOTA on a variety of molecular optimization tasks.
• Designed and implemented experiment and code tracking systems enabling facile experimentation, reproducibility and collaboration with stakeholders

Yerevan, Armenia

• Developed data acquisition protocols and worked with the design team to create an in app UI to facilitate data collection.
• Designed and coded a modular and configurable data pipeline using kedro, pandas and numpy for data cleaning, normalization, feature calculation and reporting.
• Optimized in-app data caching in swift using nuke library which reduced in-app network consumption by 40% on launch.

London, Ontario, Canada

• Validated machine learning models for prognostication of head and neck cancer patients using statistical methods
• Wrote a first-author paper detailing validation project methodology and results for journal publication
• Developed MATLAB wrapped of Python image processing software package (PyRadiomics), facilitating lab research workflow.
• Independently communicated directly with Slicer developers to resolve bugs in batch processing of CT Images.

London, Ontario, Canada

• Independently developed full bioinformatics pipeline for differential expression analyses used in downstream tasks.
• Evaluated differential protein expression from sequencing databases via meta-analyses incorporating batch normalization procedures ComBat and SVASeq.
• Optimized qPCR, immunoblot, Lowry assay and fluorescence spectrometry procedures in the Lab.