I design, build and ship AI solutions end to end: the product, the cloud infrastructure, auth, CI/CD and the security around it. Hands-on, from the first idea to production.
Agents, plugins and LLM-powered tools, built end to end and shipped to production. I own the whole path: architecture, code, cloud and deployment.
Security built into the product from the start: threat modeling, auth, secure pipelines, and automated scanning with agentic remediation. I know the trade-offs and I can explain them.
AWS and Azure with everything as code in CDK and Terraform. CI/CD on GitHub with OIDC, environments and auth. The whole path from commit to running service.
Deployed Threat Designer, the open source AI threat modeling tool from AWS, for a digital aviation company. I adapted it to enterprise security requirements, extended it with automation so an architecture description becomes a threat model, and built infrastructure around it for versioning and storing threat models.
A tool-using LLM agent built from scratch against the Anthropic API. No frameworks, raw HTTP, my own ReAct loop and tool-call protocol.
A low-level mTLS implementation running as a sidecar for zero-trust service-to-service traffic, deployed to AWS.
A small file server built to be broken. Shows path traversal in practice and how OWASP ASVS controls stop it.
I'm an AI engineer with a security background. I build AI solutions end to end and I'm hands-on in every layer: the product, the cloud, auth, the pipeline and the security around it. What I deliver is working & secure software.
Before AI, I spent multiple years as a security engineer, so I know the trade-offs and where things break.
Based in Gothenburg, happiest somewhere on the Dalmatian coast. Swedish, English, some Serbo-Croatian.
I help non-technical colleagues turn ideas into working proofs of concept, then I take them to production. I own a wide range of AI solutions end to end, from product design to implementation and security.
Deployed Threat Designer, an open source AI threat modeling tool, on AWS, adapted it to enterprise security requirements and extended it with automation. Set up SAST, SCA and SBOM tooling in GitHub with the platform team, and a secure development lifecycle measured across 100+ products.
Built a security automation framework on GitHub Advanced Security from scratch: CodeQL, secrets scanning, artifact signing, container and IaC scanning, across cloud and on-prem pipelines.
Secure boot, PKI-based key management and secure over-the-air updates for in-vehicle displays. Before that, CI/CD pipelines that cut each test cycle by about six hours.
