Rasmus Kockum
Senior Engineer & Tech Lead
I build Java backends handling 20k+ RPS, React frontends built to last, and the cloud infrastructure between them. I tune database queries, design CI/CD pipelines, and know the difference between a real microservice architecture and a distributed monolith. I have held customer search for 20 million users to a 5 ms latency ceiling on Oracle.
• Available now for advisory, architecture and review work
Three areas where I make a difference.
Senior Engineer
I write the code others avoid: high-throughput services in Java and Go, React frontends that hold up under real load, database query optimization, and infrastructure that stays up. I have tuned systems past 20k RPS and know which levers to pull, wherever in the stack they are.
CI/CD & AI Workflow Coach
I set up the pipelines, testing strategies, and review processes that let teams merge AI-generated code with confidence. In practice that means giving the agent what it needs to check itself: real documentation instead of guesswork, and a way into the running application behind the login. An agent that can verify its own work is the difference between moving fast and breaking production.
Software Architect
I design systems that stay decoupled under real-world constraints. Having seen both, I know the difference between a true microservice architecture and a distributed monolith wearing microservice clothing. I have served on architecture forums since 2016, shaping technical direction at the organizational level. I architect for EU data residency and GDPR from day one, not as a later bolt-on.
What I work with.
The tools and platforms I reach for when building systems that need to scale, stay reliable, and ship fast.
Projects I have led.
Custom platform for a CNC machine company
Acting as CTO and lead engineer for a company that manufactures and sells CNC machines. Designed and built a custom platform from the ground up: an integrated CRM, e-commerce storefront, product configurator, and order management system.
Airline service platform with full booking parity
Building a new customer service platform for SAS from scratch. A service agent can do everything a passenger can do on the SAS website — search, book, rebook, cancel, take payment — plus the things only an agent is allowed to do. Role-based access control separates agent types down to the individual action, and payments run inside the platform rather than being handed off to another system.
I designed the service boundaries, stood up the Azure infrastructure, and lead the team building it out. I run the metrics and alerting: the services autoscale, and an alert fires before the people on the floor notice.
AI agents that can verify their own work
The team writes code with AI agents. I build what makes that work: a launcher that keeps everyone on whichever model is currently best for this codebase, and an MCP server holding the entire Amadeus API documentation offline — searchable, and formatted to be read by an agent rather than by a human. The agent reads the real API instead of guessing at it.
It can also check its own work. Amadeus integrations verify automatically, and MSAL authentication is solved end to end with Playwright, so an agent can log in through SSO and confirm the frontend behaves before a human opens it.
Cloud native Passenger Service System
Led development of a new PSS — the reservation core an airline runs on. Every online booking SAS takes is created through it. Designed the cloud native architecture on Azure, defined the CI/CD pipeline, and drove the team to meet strict airline industry uptime and latency standards. Metrics, alerting and autoscaling are mine.
Airline alliance migration in 8 months
Switched a Nordic carrier from one global alliance to another, end to end. Coordinated across reservation systems, departure control, and passenger handling on a tight deadline with zero room for downtime.
Stock portfolio recommendation platform
Architected, built and deployed a platform for stock portfolio recommendations. Implemented real-time stock price updates with websockets and a React frontend.
Customer search at 20M users, capped at 5ms
Led the team building a customer search system serving 20 million users with a hard latency ceiling of 5 milliseconds. Tuned the Oracle data model, indexing, and query paths to keep worst-case response times flat under load.
Core banking migration to SAP
Led migration from a legacy core banking system to SAP, coordinating across multiple teams in the organization. Managed the technical integration work while keeping existing operations running throughout the transition.
Breaking out a monolith, piece by piece
Led the first extraction from a large monolithic application into independent services. The approach proved itself: subsequent extractions followed across multiple teams, cutting deployment cycles and enabling parallel feature development.
Selected writing.
- May 31, 2026
Shipping AI-Generated Code with Confidence
The question with AI is no longer whether an agent can write the code. It's whether you can ship it. Here are the automated gates that let a team merge AI-written features without breaking production: OpenAPI breaking-change detection, SpotBugs, enforced coverage, and integration tests recorded from real third-party calls.
- December 17, 2024
The Perils of Premature Abstraction: When Code Duplication Isn't a Problem
Understanding why abstracting and centralizing duplicated code too early can be restrictive and how to avoid premature abstraction.
- July 26, 2024
Maximizing application performance by optimizing HTTP and Database operations
Understanding why reducing and optimizing HTTP and Database calls is the most efficient way to improve application performance, with practical tips and examples.
Interested in working together?
I'm available now for advisory, architecture, and review work, and open to full-time engagements from Q1 2027. The best way to reach me is email.