-
for Science @ Scale: Pretraining, instruction tuning, continued pretraining, Mixture-of-Experts; distributed training/inference (FSDP, DeepSpeed, Megatron-LM, tensor/sequence parallelism); scalable evaluation
-
, enhanced sampling, QM/MM) Experience improving performance and scalability of simulation workflows via: Parallelization and performance engineering GPU/accelerator optimization Algorithmic innovation
-
for Science @ Scale: Pretraining, instruction tuning, continued pretraining, Mixture-of-Experts; distributed training/inference (FSDP, DeepSpeed, Megatron-LM, tensor/sequence parallelism); scalable evaluation
-
of simulation software to model complex system flows in processes relevant to the nuclear nonproliferation mission space. You will be expected to be a heavy end user of the tools generated, participate in model
-
orchestration (e.g., Airflow, Prefect, Dagster, Make/Snakemake) and working in Linux/HPC environments. Experience with large, multi-resolution geospatial datasets and performance-oriented processing (tiling