echoloc
Get started
echoloc/ Companies/ Firmus Technologies

Firmus Technologies Tech Stack

AI infrastructure operator optimizing data centers for training efficiency

Technology, Information and Internet St Leonards, Tasmania 51–200 employees Privately Held
firmus.co LinkedIn

Firmus Technologies operates physical AI infrastructure—data centers built around liquid cooling and GPU orchestration to maximize training throughput per unit of energy cost. The stack reflects a deep systems engineering org: Kubernetes, Slurm, and Megatron-LM for workload scheduling; PyTorch, JAX, and DeepSpeed for training; and low-level hardware telemetry (InfiniBand, Netbox, Cumulus Linux) across Dell, HPE, and Supermicro servers. Engineering-heavy hiring (13 roles open, mostly senior and mid-level) paired with active data center construction and GPU deployment projects signals rapid scaling of physical capacity.

Tech Stack 77 technologies

Core StackKubernetes PyTorch Jenkins ArgoCD GitLab CI/CD Python Go Ansible Helm AWS Cisco Slurm Megatron-LM JAX DeepSpeed NCCL LoRA QLoRA Bash Netbox Cumulus Linux Sonic InfiniBand CWDM Azure Dell HPE Supermicro Arista Juniper+47 more

What Firmus Technologies Is Building

◆Challenges

  • Ai energy challenge
  • Maintaining hpc infrastructure
  • Energy-efficient ai infrastructure
  • Connecting ai workload performance to energy efficiency
  • High cost of ai training
  • Low-cost ai tokens
  • Scaling ai compute
  • Optimising time to first token
  • Minimising downtime
  • Optimizing training efficiency

▲Active Projects

  • Sustainable ai factory development
  • Ai factory construction
  • Ai infrastructure network deployment
  • Data center fit-out projects
  • Kubernetes cluster deployment
  • Gpu server deployment
  • Ai-accelerated hpc infrastructure maintenance
  • Onboarding and migration programs
  • Ai factory developments
  • Production-ready training recipes

Hiring Activity

Accelerating25 roles · 20 in 30d

Department

Engineering
13
Ops
6
Finance
3
Construction
1
Procurement
1
Security
1
Support
1

Seniority

Senior
12
Mid
10
Manager
2
Director
1
Lead
1

Notable leadership hires: Cost Lead

Company intelligence

Find more companies like Firmus Technologies by tech stack, pain points and active projects

Get started free

About Firmus Technologies

Firmus Technologies designs and operates AI data centers optimized for training workload efficiency and profitability. The company's core product—what they call AI Factories—combines advanced liquid cooling systems, GPU cluster orchestration, and energy management across the full stack from power grid integration to networking telemetry. They are actively constructing new facilities in Australia, with hiring expansion underway in the United States and Singapore. The company targets AI model-training providers and compute platforms seeking lower token-generation costs through infrastructure efficiency rather than hardware alone.

HeadquartersSt Leonards, Tasmania
Company Size51–200 employees
Hiring MarketsUnited States, Singapore, Australia

Frequently Asked Questions

What is Firmus Technologies' tech stack?

Kubernetes, Slurm, and PyTorch for orchestration and training; Megatron-LM, DeepSpeed, and JAX for model optimization; InfiniBand and Cumulus Linux for networking; Ansible and ArgoCD for infrastructure automation; plus HPE, Dell, and Supermicro hardware platforms.

What is Firmus Technologies building?

Data center construction projects, Kubernetes cluster and GPU server deployments, AI infrastructure network expansion, and production-ready training recipes. Current focus includes sustainable AI factory development and migration programs for new facilities.

Similar Companies in Technology, Information and Internet

Other companies in the same industry, closest in size

F
Fluidstack
Technology, Information and Internet

How this profile is built

Firmus Technologies's technology stack, projects, and hiring signals are inferred from public hiring and company data — career pages, public listings, and company web presence — then clustered and de-duplicated. Figures are estimates that refresh over time. Read our full methodology →

This is not an official vendor or customer list. It is a technology-adoption signal inferred from public data, intended for B2B research.