About me

Hi! I am Kiarash Jamali, a machine learning researcher currently applying it to structural biology. I got my B.S.c in mathematics and statistics from the University of Toronto, St George in 2021. While in Toronto, I was at the Vector Institute as a machine learning researcher being supervised by Frank Rudzicz. While there I worked mostly on developing new neural network architectures, specifically for GANs and for metrics.

In the summer of 2019, as part of UofT’s PEY program, I had a lot of fun working as a machine learning researcher at RBC Capital Markets on the Aiden project. Aiden is a first-of-its-kind reinforcement learning agent that executes large equities orders on behalf of RBC’s clients in the market while minimizing the slippage that is the consequence of such large orders. I built different reinforcement learning models while also greatly improving the infrastructure of the pipeline.

After graduating in the winter of 2020, I started working for Semantic Health as a machine learning scientist. Semantic Health is a start-up that uses NLP to help hospitals maximize the use of their data by medical coders and auditors. I worked on localization algorithms using vector similarity search as well as more conventional language modelling tasks with transformers. I also set up the machine learning productionization pipeline with MLFlow.

Since October of 2021, I have been pursuing a PhD at the MRC Laboratory of Molecular Biology under the supervision of Sjors Scheres and funded by the LMB Cambridge Scholarship. I have been working on applying deep learning to cryogenic electron microscopy (cryo-EM). My main project has been focused on automated atomic model building in cryo-EM maps. I have also been working on heterogeneous reconstruction algorithms using VAEs.

I spent the summer of 2023 as a machine learning intern at Diffuse Bio, where I was able to learn from Namrata Anand about machine learning approaches to protein design.

I have also made some open source contributions to PyTorch and e3nn-jax.

If you would like to learn more, or have an interesting cryo-EM structure that you would like to test with my auto-building, please reach out via email.