Deloitte built its Quartz Atlas AI drug discovery platform on NVIDIA BioNeMo, DGX Cloud and NIM microservices.
Deloitte needed to integrate fragmented multimodal data from dozens of sources and apply advanced AI models to preclinical validation. Quartz Atlas AI combines BioNeMo structure-prediction models (AlphaFold2, OpenFold, ESMFold), a fine-tuned 15B-parameter ESM2 model trained on DGX Cloud, MolMIM and MegaMolBART for molecule generation, and a knowledge graph of 12 million nodes and 97 million edges. The page reports a 50% gain in developer productivity, 7-10 months saved on multi-node training setup, training cut from 4 weeks to 8 hours, and pipeline assembly cut from 4-6 weeks to a few clicks.
NVIDIA is a computing and AI company whose BioNeMo and Clara platforms provide models and accelerated tools for drug discovery, genomics and medical imaging.
This sits in the Build stage of the AI use-case lifecycle: Named examples from the foundation-model labs and platforms pharma teams use to actually construct, train and generate the model or application.
Published by NVIDIA on its own site — read the original case study on nvidia.com ↗
Related use cases
- 01Amgen trained antibody language models on NVIDIA BioNeMo and DGX Cloud, reaching three-month goals in four weeks.
- 02Biofy cut infectious disease diagnosis from five days to four hours with an NVIDIA-accelerated genome pipeline.
- 03A nonprofit built a medical knowledge LLM platform on NVIDIA NIM for 17,000+ practitioners in 19 countries.
- 04Astellas trained its own antibody language models on NVIDIA BioNeMo to predict antibody properties before measurement.
- 05OpenAI's dedicated healthcare vertical page/launch post.
- 06Frontier reasoning model built for biology, drug discovery and translational medicine; named qualified customers include Amgen, Moderna, Allen Institute, and Thermo Fisher Scientific.