I'm Michael Hopwood, data scientist at Microsoft. When I was pursuing my PhD, I was focused on graph neural networks and probabilistic machine learning especially in their applications to physics. Since then, I have dropped out to pursue work in industry. For a list of my publications see my GoogleScholar or you can review my CV.
News & Updates: (Click to expand)
- July 2023. Started as data scientist at Microsoft.
- May 2023. Dropped out of PhD, graduating with Master's degree in Data Science and Statistics
- August 2022. Began internship on Amazon's risk analysis team working on graph neural networks applied science.
- May 2022. Began internship on Microsoft's bing search optimization team working on optimal loss functions and efficient productionization.
- January 2022. Began internship on Tesla's charging data modeling team working on network optimization and timeseries modeling.
- January 2022. Passed master's comprehensive exam in Data Science.
- October 2021. Oral presentation at INFORMS Annual Meeting 2021 regarding a failure detection technique using gaussian-emission hidden markov models
- August 2021. Invited to speak at network science conference, ICUFN 2021 about work ( proceedings ) which validated active learning practices with simulations (an extension from the previous journal paper).
- May 2021. Released open-source python package tackling machine learning & simulation applications in photovoltaic systems.
- April 2021. Journal paper published which explores a phenomenon that ties network topology to active learning in graph neural networks
- April 2021. Participated in stanford datathon and submitted report about applications of generalized low rank models to garage parking capacity
- March 2021. Won 2nd place in 2021 OUC Data Science Competition focused on Electric Vehicle Detection
- December 2020. Presented at AGU a methodology using data fusion techniques (both NLP and timeseries) to study the effect of extreme weather events on photovoltaic systems
- September 2020. Journal paper published studying the use of neural networks on failure classification in PV systems
- August 2020. Began company which designed, built, and deployed a Bayesian ML-informed algotrading agent, using the funds of an angel investor, along with two other software developers.
- August 2020. Started my PhD at UCF!
- June 2020. Presented at IEEE PVSC 47 (and won best student paper )about the use of principal component analysis and random forest (RF) on current-voltage curves in a failure classification task; released in a paper
- May 2020. Began R&D internship at Sandia National Labs!
- August 2019. Release first open-source machine learning package using physics-informed kernels and unsupervised learning focused on energy modeling in photovoltaics systems which has, to date, over 6k downloads.
- June 2019. Project accepted to IEEE PVSC 46 delineating methods of physically simulating failures in PV systems
For more updates, please visit my personal page.