Teams at Proxima

Optimize training and inference performance for structural and generative ML models. Develop custom CUDA and Triton kernels, use ML compilers, scale distributed training across multi-node GPU clusters, reduce inference costs, improve GPU utilization on GCP, and build benchmarking and profiling tools. Provide technical direction, select infrastructure, influence research teams, and mentor engineers.

Lead research and design of large-scale deep learning systems for protein structure prediction and molecular glue design. Develop novel architectures and training paradigms, collaborate with chemistry/biology experts, publish at top ML venues, and apply models to drive drug discovery value.