Jon Thor K

Lead HPC/Grid Engineer

Millennium Management • Jul 2023 - Present • London, UK (Hybrid)

• • Leading enterprise-scale HPC platform transformation to support quantitative trading teams, portfolio managers, developers, and data scientists
• • Architecting hybrid cloud HPC infrastructure seamlessly integrating on-premises and public cloud resources using Slurm scheduling at scale
• • Developed comprehensive automation framework using Ansible playbooks with custom APIs for dynamic inventory management
• • Built full-stack self-service platform with Go-based APIs/CLIs and React web interface for on-demand cluster provisioning
• • Created advanced tooling ecosystem for workload scheduling optimization and performance analysis across trading operations

Interim Product Manager of MAAS / Product Manager of HPC

Canonical • Feb 2022 - Jul 2023 • London, UK

• As Product Manager of HPC (Feb 2022 - Jul 2023):
• • Reported directly to CEO Mark Shuttleworth to lead Canonical's HPC strategy
• • Leveraged Ubuntu's 14% HPC market share and dominance in AI training workloads
• • Forged strategic partnership with NVIDIA for InfiniBand optimization on Ubuntu
• • Authored industry-recognized reference architectures with leading OEMs
• • Built thought leadership through presentations at Open Infrastructure Summit, ARM Dev Summit, NASA Cybersecurity Day
• As Interim Product Manager of MAAS (Apr 2022 - Jul 2023):
• • Defined strategic roadmap for Metal as a Service focusing on HPC cluster provisioning
• • Architected critical features: diskless booting, InfiniBand support, cluster delivery
• • Established five-year product vision influencing current development cycles

CTO HPCFLOW

atNorth • Jan 2018 - Dec 2021 • England, UK (Contract)

• • Scaled HPCFLOW platform from single to multi-regional HPC-as-a-Service offering
• • Developed enterprise-grade HPCaaS with Intel Select Solutions and Open HPC
• • Led pre-sales and solution architecture for enterprise HPC requirements
• • Pioneered Omni Path vFabric multi-tenancy integrated with OpenStack Neutron
• • Engineered custom networking for HPE Flex Fabric and Cumulus OS environments
• • Presented at ISC HPC and Supercomputing conferences

Senior Solution Architect HPC

Advania • May 2016 - Jan 2018 • Reykjavík, Iceland

• • Architected and built HPCFLOW IaaS platform for HPC/AI workloads from ground up
• • Integrated OpenStack, Ironic, Packer, Slurm, and Ceph for HPC clusters as service
• • Led pre-sales technical consulting enabling cloud migration at on-premises performance
• • Pioneered early Kubernetes adoption (v1.0 in 2016) for internal services
• • Leveraged HPE Performance Cluster Manager for automated HPC deployment

Senior Deployment Engineer / Core Developer

GreenQloud • Dec 2013 - Apr 2016 • Reykjavík, Iceland

• As Senior Deployment Engineer (Aug 2014 - Apr 2016):
• • Architected cloud infrastructure using Chef automation and Apache CloudStack
• • Pioneered automated deployment for private cloud components
• • Developed bare metal provisioning for CloudStack with multi-tenant networking
• As Core Developer (Dec 2013 - Aug 2014):
• • Extended Apache CloudStack core capabilities using Java
• • Developed custom features expanding enterprise platform capabilities

Programmer

Destino • Jun 2011 - May 2013 • Stavanger, Norway

• • Developed enterprise content management system using C# and JavaScript
• • Introduced SCRUM methodology and continuous integration workflows
• • Built full-stack web applications with robust performance
• • Established CI/CD pipelines automating testing and deployment

Charmed Operator Day • Technical Session

In this session we will introduce MAAS. How charms can be used with MAAS to deploy anything from LXD to k8s and OpenStack. And how that can be extended to deploy bare metal workloads such as those that exist in the realm of HPC straight to a cluster of nodes all through MAAS using Juju and charms.

OpenStack for Computational workloads

OpenInfraSummit • Technical Presentation

Computational workloads such as High Performance Computing (HPC), High Performance Data Analytics (HPDA) and Machine Learning (ML) workloads are increasingly demanding more flexibility and agility in delivery and overall ease of use, while still calling for optimal performance and scale.