We don’t just do AI. We do AI that works.
Booth #184
The BFSI industry is in a perfect storm, with accelerating forces transforming every financial value chain. Whether you’re an established insurer, a big bank or a fintech scale-up, you don’t need more AI experiments — you need an enterprise that is AI-ready.
To grow organically, your foundation must evolve. At Thoughtworks, we pair deep BFSI industry knowledge with AI/works™, our agentic development platform, to move you from legacy constraints to agentic architectures in 90 days.
Visit our booth in the Financial Services Pavilion to help navigate the disruption:
Agentic Transformation with our unified control plane: Move from disconnected AI experiments to governed, scalable agentic systems.
Modernizing Legacy Systems with AI-First Software Delivery: Accelerate modernization of legacy platforms and mainframes with AI-driven engineering.
Advanced Fraud Detection with Thoughtworks and NVIDIA: Improve fraud detection accuracy while reducing costly false positives.
We hope you can join us to explore the next wave of AI innovation.
Our session
From Lists to Graphs: Live Demo of Graph Neural Networks for Real-Time Fraud Detection
Fraudsters think in networks — yet most fraud systems still think in rows and columns. In this live demo session, we move beyond theory to show how Graph Neural Networks (GNNs) fundamentally change fraud detection by modeling relationships, not just transactions. Using NVIDIA’s AI Blueprint for fraud detection, we demonstrate how graph-based intelligence exposes coordinated fraud rings, synthetic identities, and money mule networks that remain invisible to traditional tabular machine learning models.
Attendees will see an end-to-end, production-grade fraud inference pipeline in action:
- Streaming transactions ingested in real time
- Subgraph extraction from a live financial graph
- GPU-accelerated GNN inference generating relational risk embeddings
- Hybrid scoring with XGBoost for fast, explainable decisions
- Sub-150ms decision latency suitable for instant payments
- We will show how GNN embeddings act as a network intelligence layer — augmenting existing fraud models rather than replacing them — while preserving regulatory explainability through SHAP analysis and graph-level attribution.
This session is ideal for data scientists, ML engineers, fraud platform architects, and security engineers who want a practical, hands-on understanding of how graph-based AI works in real financial environments.
Visit us on Tuesday, March 17th from 5:00 - 7:00 pm at the AI Competency Kiosk (#5) in the AWS booth (#921).
