Machine Learning Engineer
+ Led a team to develop a stochastic variational bayesian inference and supervised machine learning infrastructure for various classification tasks, using scalable ETL processes to integrate structured data (HIPAA Compliance), successfully bayesian-search trained billions of parameters at 3.1X optimized compressed RAM (2 Terabytes/node) and achieved ≥87% prediction accuracy+ Launched and developed several distributed solutions for the clients using Neural ODE training and paralleled Neural architecture search on QuadroP4000 GPUs and X86 Xeon servers (>40 TFLOPs, O(1) memory O(E) time), achieved 4X speedup and deployed at heterogeneous VM & Container infrastructures + Built software pipelines for deep neural network architecture search and CNN training, fed data via CUDA GPUDirect RDMA, then trained them on proprietary Keras, Tensorflow, and TFX built, achieving 1.5X throughput speedup
