Shivin D.
About Shivin D.
Shivin D. is a Machine Learning Engineer II at Singular Genomics in San Diego, California, where he has worked since 2021. He holds a Bachelor's and a Master's degree in Electrical Engineering and Computer Science from UC Berkeley, and has made significant contributions to machine learning efficiency and DNA sequencing accuracy.
Work at Singular Genomics
Shivin D. has been employed at Singular Genomics as a Machine Learning Engineer II since 2021. In this role, he has automated and optimized the entire machine learning training and deployment pipeline, achieving a fourfold increase in model training speed. His contributions include the implementation of a stateful machine learning model architecture that enhanced the accuracy of DNA sequencing machines, significantly reducing calibration errors and incorrect reads.
Education and Expertise
Shivin D. earned a Bachelor of Science in Electrical Engineering and Computer Science with Highest Honors from UC Berkeley College of Engineering, completing his studies from 2017 to 2020. He further pursued a Master's of Science in the same field at UC Berkeley, studying from 2020 to 2021. His educational background is complemented by practical experience as a tutor and teaching assistant at UC Berkeley, where he supported courses in electrical engineering.
Background in Research and Teaching
Prior to his current role, Shivin D. worked as a researcher at Berkeley Artificial Intelligence Research from 2019 to 2021. He also served as a tutor for EE16A and EE123 courses, as well as a teaching assistant for EE127 Optimization Models in Engineering. His teaching roles spanned from 2018 to 2020, providing him with a solid foundation in both academic support and research.
Technical Achievements
Shivin D. has implemented several significant technical improvements throughout his career. He achieved a 16x reduction in RAM usage and improved IO speed by encoding data in int8 format. Additionally, he implemented data compression techniques that resulted in a 38% lossless compression, saving $120,000 annually in AWS S3 storage costs. His work with the NVIDIA Triton Inference Server led to a 2.93x speedup in GPU inference.
Optimization and Performance Improvements
Shivin D. has demonstrated expertise in optimizing machine learning processes. He achieved a 3.3x speedup in CPU time by enhancing data processing, reducing runtime from 24 hours to 17 hours. His efforts in improving machine learning pipelines and model performance have had a notable impact on operational efficiency.