The Tech Stack Powering Wise

2025-12-15 · Source: ByteByteGo Newsletter · Field: Technology & Digital — Software Development & Engineering, Cloud Computing & IT Infrastructure, Artificial Intelligence & Machine Learning · Depth: Intermediate, long

Summary

Wise, a financial technology company processing £36 billion quarterly, manages its extensive engineering operations with over 850 autonomous engineers across 1000+ microservices by treating its internal infrastructure as a product. This approach standardizes development and deployment through a microservice chassis framework, versioned artifacts for dependency management, and a Compute Runtime Platform (CRP) built on Kubernetes. Their deployment system, Spinnaker, automatically blocks incidents by routing 5% of traffic to new versions, monitoring metrics for 30 minutes, and rolling back anomalies. Wise also uses a unified observability stack (Loki, Grafana, Tempo, Mimir) to correlate logs, traces, and metrics, and has implemented a secure gateway for LLM providers like Anthropic and Google Gemini, alongside a custom RAG service for internal context.

Key takeaway

For AI Architects and MLOps Engineers building complex, distributed systems, Wise's strategy of treating internal infrastructure as a product offers a blueprint for scaling autonomous teams while maintaining reliability. You should evaluate adopting versioned chassis frameworks and automated canary deployment systems like Spinnaker to standardize development and deployment, significantly reducing incidents and engineering overhead.

Key insights

Standardized internal platforms enable autonomous teams to scale complex, high-reliability financial services.

Principles

• Treat internal infrastructure as a product.
• Standardize starting points and deployment paths.
• Automate dependency and security updates.

Method

Wise employs a microservice chassis as a versioned artifact, not a template, ensuring updates flow downstream. It uses Spinnaker for canary deployments with automated metric-based rollbacks and a unified LGTM observability stack for incident correlation.

In practice

• Implement a versioned microservice chassis.
• Adopt automated canary deployments with rollbacks.
• Consolidate observability tools for correlation.

Topics

• Internal Platform Engineering
• Microservice Architecture
• Automated Deployments
• Kubernetes Infrastructure
• Data Engineering

Best for: MLOps Engineer, Software Engineer, AI Architect

Related on AIssential

• AI made your app teams 10x faster. Nobody gave your platform team 10x the headcount. — OpenAI's data platform team manages all core data systems, from analytics to ML infrastructure.
• Mini book: Architecting Autonomy: Decentralising Architecture Inside an Organization — Decentralized architecture with automated guardrails enables organizational scale and velocity in the AI era.
• Autonomous Drone Delivery at Scale — Scaling autonomous drone delivery demands vertical integration, custom software, and a "cattle, not snowflakes" approach to fleet management.
• How Pinterest Built a Production MCP Ecosystem — Standardized protocols and robust platform infrastructure are crucial for scaling AI agent integration with internal systems.
• The Product Engineer Company: How Portkey Works and Builds Its Product — Portkey's success stems from deep product ownership by engineers and a disciplined focus on scalable AI infrastructure.
• HPE expands self-driving networking strategy as AI moves into production — HPE is positioning itself as a full-stack AI infrastructure provider, emphasizing self-driving networks and AI factories for agentic AI deployment.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by ByteByteGo Newsletter.