Andreas Jansson is a Principal Systems Engineer with 15 years of experience building scalable ML and infrastructure systems, now at Cloudflare after cofounding and serving as CTO of Replicate. He blends deep machine-learning tooling expertise with production-grade backend and DevOps skills, contributing to notable open-source projects like replicate/cog (containerized ML runtimes) and keepsake (model versioning and storage backends). His background includes staff ML engineering at Spotify and early systems roles in music tech, underpinned by a PhD in Music Informatics and formal training in audio and electronic music production. Andreas is comfortable moving models from research to cloud-native deployment, implementing storage backends, CI/CD, and testing at scale. He pairs technical leadership with hands-on Go, Python, and cloud service work, and brings a musician’s perspective to signal and data problems that often uncovers unconventional but practical solutions. Based in Sweden, he focuses on reliable ML infrastructure that bridges researcher workflows and production requirements.
15 years of coding experience
17 years of employment as a software developer
Audio Engineering Diploma, Audio Engineering, Audio Engineering Diploma, Audio Engineering at SAE London
PhD, Music Informatics, PhD, Music Informatics at City St George’s, University of London
Electronic Music Production, Electronic Music Production at SAE Stockholm
Contributions:1 release, 89 reviews, 235 commits in 1 year 7 months
Contributions summary:Andreas primarily focused on developing backend infrastructure, contributing code for cloud build processes and server functionality using Go. They worked on setting up and utilizing cloud services like Google Cloud Storage (GCS) and Google Cloud Build. The commits also show interaction with Docker, indicating a focus on containerization and deployment aspects of the project.
Contributions:1 release, 76 reviews, 211 commits in 8 months
Contributions summary:Andreas implemented S3 and GCS storage backends for machine learning model version control, using boto3 and aiohttp. They added unit and integration tests for the new backends using pytest and moto. The user also designed the CLI scaffolding and added functionality for managing experiments, including creating, listing, and deleting them, along with implementing features for the model training process.
machine-learningversion-control
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.