A PhD in Snapshots
2 days ago
- #Machine Learning
- #PhD Journey
- #Drug Discovery
- A PhD typically takes 5-6 years, involving coursework, research rotations, and an apprenticeship-like relationship with a faculty advisor.
- The author's PhD journey was supported by the Hertz Foundation, requiring bi-annual progress reports to demonstrate research progress.
- Initial years involved coursework in machine learning and algorithmic design, followed by research rotations without finding a perfect fit.
- The author shifted focus to computational biology, joining the Pande group to apply machine learning to protein simulations.
- Key projects included developing deep learning systems for drug discovery, creating the open-source package DeepChem, and curating datasets like MoleculeNet.
- Collaborations with Google and pharmaceutical companies like Pfizer and Merck were significant, focusing on improving drug discovery processes.
- The author faced challenges with paper rejections from machine learning venues due to a focus on biological applications, leading to submissions in biochemical journals.
- Mentoring younger students and organizing the first deep-learning in chemistry conference were highlights of the later PhD years.
- The PhD concluded with the defense in December 2017, followed by plans to co-found a startup in the blockchain space.