Real-time Analytics
NeoSpace needed to make tabular foundation models viable for financial institutions that serve tens of millions of customers, but time-to-train, inference latency, and the cost of running transformer workloads on traditional ML infrastructure blocked real-time decisioning at production scale.
Reported impact
NeoSpace reported 30% to 50% higher model accuracy than traditional ML, a 19% lift versus its own baseline after more training iterations, and a 10% increase in installment credit offers at a major Latin American bank while improving hiring and conversion rates in the top scoring decile.
NeoSpace built its NeoData platform on NVIDIA accelerated computing so banks can run foundation models over trillions of tabular, event, and unstructured records. In production on Oracle Cloud Infrastructure, NeoSpace uses NVIDIA GB200 NVL72 systems and the CUDA software stack to train and serve domain models for tasks such as credit propensity scoring. Its platform combines CUDA Toolkit, NCCL, NVSHMEM, GPUDirect, cuBLAS, CUTLASS, CuTe, RAPIDS, TensorRT LLM, and Dask, while an internal orchestration layer called NeoCore runs directly on bare-metal hosts to distribute workloads across multi-cloud environments. The architecture lets NeoSpace replace fragmented ML pipelines with a single predictive layer that can ingest enterprise feature stores, historical transactions, and customer events and then return real-time scores for underwriting and marketing decisions.
Enterprise feature stores, eligibility variables, transaction records, and behavioral history are processed through NeoData on OCI; RAPIDS and CUDA-accelerated jobs prepare and train the models; NeoCore orchestrates serving, and TensorRT LLM returns real-time predictions that banks use for credit and marketing decisions.
10% increase in installment credit offers.
30%–50% higher model accuracy compared with traditional ML approaches.
NeoData now operates at scale across a multi-cloud environment for one of Latin America's largest private banks with more than 60 million customers.
NeoSpace is among the first companies in Latin America to deploy NVIDIA GB200 NVL72 systems in production, through Oracle Cloud Infrastructure (OCI).
Review how this reference was scored and trace each normalized claim back to its documented source.
Scoring methodology
Customer Reference Value measures usefulness as a customer reference. Evidence Confidence measures the quality and traceability of available evidence, not independent verification.
Traceability
Field coverage shows which facts each source supports. Dates are shown only when documented by the source.
Related
NVIDIA Sample Aug 12, 2026 CYRAN AI Solutions CYRAN needed to deliver interactive geospatial intelligence in air-gapped and sovereign environments, but JPEG 2000 decode latency on large satellite scenes was the bottleneck that kept edge analytics from being… 26 GB satellite scene NVIDIA DGX Spark · NVIDIA nvJPEG2000 · PixelFlux NVIDIA Sample Jan 12, 2026 XPAI XPAI needed a privacy-preserving way to measure how people use and perceive physical spaces so brands and event operators could optimize offline experiences with real-time evidence instead of intuition. millions of data points per second AVAI One Sensor · NVIDIA DeepStream SDK · NVIDIA Jetson Orin 
Google Cloud Sample Archived · Oct 16, 2026 Alta Alta needed a transactable Google Cloud Marketplace path for its AI and technology spend management platform without pulling its product team off core forecasting work, while also helping finance and IT leaders… Saved 125 engineering and 60 project management hours Cloud Pub/Sub · Cloud Run · Cloud SQL 