Wealthsimple needed a standardized way to deploy and serve machine-learning models for fraud, transaction routing, onboarding, and personalization, but framework-specific inference stacks were taking months to ship and causing outages that hurt customer transfers.
Reported impact
Wealthsimple reduced model delivery time from months to under 15 minutes, raised inference uptime from 95 percent to 99.999 percent, cut serving latency by 20 percent, and delivered 145 million predictions over 12 months with no IT tickets tied to AI inferencing.
Wealthsimple turned its fragmented model-serving environment into an internal ML-as-a-service platform by standardizing on NVIDIA Triton Inference Server and GPU-backed deployment on AWS. Triton replaced bespoke, framework-specific servers with a single inference layer that could run on both CPUs and GPUs, while A10G instances and the NVIDIA GPU-Optimized AWS Machine Image gave the company a repeatable runtime for production serving. On top of that platform, Wealthsimple now runs more than 30 models across fraud detection, suspicious-transaction analysis, customer recommendations, onboarding, and institutional-transfer routing. Because the serving layer is standardized, model developers can deploy without waiting on a separate engineering team, and the platform organization can focus on shared reliability and cost tuning instead of one-off integrations for each model family.
Feature data from onboarding, transaction, and recommendation workflows flows into models served by Triton on A10G-backed AWS environments. Predictions are returned to Wealthsimple's product systems for fraud detection, transfer routing, and personalization, while the shared platform lets new models move into production without bespoke serving work.
Following the change, model delivery time has been slashed from months to under 15 minutes-a game-changing advancement for the organization.
Today, Wealthsimple runs models powered by NVIDIA A10G GPUs on the AWS cloud, tapping into the NVIDIA GPU-Optimized AWS Machine Image to drive efficiency and innovation.
Today, Wealthsimple, which manages over C$30 billion in assets, supports over 30 AI models that generated over 145 million predictions in the past 12 months.
Triton Inference Server changed the game, delivering an impressive 99.999% uptime.
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 Jun 2, 2026 Revolut Revolut needed to replace siloed fraud, credit, marketing, and recommendation models that depended on months of feature engineering with a single behavioral foundation model that could learn from raw financial event… 16.1% improvement in credit risk prediction performance from NVIDIA Llama-Nemotron-Embed-1B-v2 NVIDIA H100 GPUs · NVIDIA Llama-Nemotron-Embed-1B-v2 · Nebius AI Cloud NVIDIA Sample Mar 17, 2026 Mastercard Mastercard wants to turn its transaction data into a reusable AI insights engine for payments and commerce, especially to strengthen cybersecurity and fraud detection, improve personalization and loyalty use cases, and… billions of transactions Databricks · Nvidia's advanced accelerated computing platform · large tabular model (LTM) NVIDIA Sample Feb 2, 2026 RBC Capital Markets RBC Capital Markets needed to speed up document-heavy research and banking workflows so analysts and bankers could deliver faster insights, prepare client materials more efficiently, and answer complex questions without… 10x increase in document processing capacity NVIDIA AI Enterprise · NVIDIA NIM · NVIDIA NeMo 