Eric Dawson

Background

About Eric Dawson

Cambridge Genetics PhD working on inference acceleration of life sciences AI models at NVIDIA. I currently lead development of BioNeMo protein structure prediction NIMs for running AlphaFold2, Boltz-2, and MMseqs2 at massive scale. I am also head architect of the BioNeMo Framework single-cell data loader (SCDL) pip package, designed to handle billions of cells efficiently during training and inference, and former Parabricks contributor. I focus on unraveling biological mysteries and enabling other researchers to solve their intractable problems using accelerated computing. At the National Cancer Institute I led the largest genomic study to date of Chernobyl Tissue Bank samples, using cloud computing to analyze more than 400 deep-sequenced WGS samples during my PhD. I contributed early methods for structural variation calling on variation graphs to the Variation Graph (vg) Toolkit and developed a method for detecting HPV coinfections 100x faster than the previous state of the art. As an undergrad I contributed to building, supporting, and optimizing applications for supercomputing infrastructure (including the fourth-fastest system on the TOP500) and was the first biology student to compete in, lead a team in, and win the Student Cluster Competition. I am a biologist, futurist, HPC performance engineer, and big data scientist. I have worked on viral, plant, animal, human, and cancer genomes and pangenomes, and I teach NGS Bioinformatics and Cancer Genome Analysis as part of the Wellcome Overseas Courses. I am committed to making genomic analysis practically free, seemingly instant, and accessible to every human being within my lifetime.