Asks for

The ad says“Proven experience optimising large-scale AI/ML systems in production. You have shipped real improvements, not just prototypes.”
The ad says“Experience with cloud AI infrastructure on AWS, GCP or Azure, and with inference serving frameworks such as vLLM, TGI or TorchServe.”
The ad says“Experience with cloud AI infrastructure on AWS, GCP or Azure, and with inference serving frameworks such as vLLM, TGI or TorchServe.”
The ad says“Hands-on knowledge of LLMs, transformer architectures and the trade-offs involved in deployment, including quantisation, distillation, batching, caching and RAG.”
The ad says“Hands-on knowledge of LLMs, transformer architectures and the trade-offs involved in deployment, including quantisation, distillation, batching, caching and RAG.”
The ad says“Hands-on knowledge of LLMs, transformer architectures and the trade-offs involved in deployment, including quantisation, distillation, batching, caching and RAG.”
The ad says“Experience with cloud AI infrastructure on AWS, GCP or Azure, and with inference serving frameworks such as vLLM, TGI or TorchServe.”
The ad says“Hands-on knowledge of LLMs, transformer architectures and the trade-offs involved in deployment, including quantisation, distillation, batching, caching and RAG.”
The ad says“Strong software engineering fundamentals. Python proficiency is a given and familiarity with MLOps tooling such as MLflow, Weights & Biases or Ray is expected.”
The ad says“Strong software engineering fundamentals. Python proficiency is a given and familiarity with MLOps tooling such as MLflow, Weights & Biases or Ray is expected.”
The ad says“Strong software engineering fundamentals. Python proficiency is a given and familiarity with MLOps tooling such as MLflow, Weights & Biases or Ray is expected.”
The ad says“Hands-on knowledge of LLMs, transformer architectures and the trade-offs involved in deployment, including quantisation, distillation, batching, caching and RAG.”
The ad says“Hands-on knowledge of LLMs, transformer architectures and the trade-offs involved in deployment, including quantisation, distillation, batching, caching and RAG.”
The ad says“Strong software engineering fundamentals. Python proficiency is a given and familiarity with MLOps tooling such as MLflow, Weights & Biases or Ray is expected.”
The ad says“Experience with cloud AI infrastructure on AWS, GCP or Azure, and with inference serving frameworks such as vLLM, TGI or TorchServe.”
The ad says“Experience with cloud AI infrastructure on AWS, GCP or Azure, and with inference serving frameworks such as vLLM, TGI or TorchServe.”
The ad says“Hands-on knowledge of LLMs, transformer architectures and the trade-offs involved in deployment, including quantisation, distillation, batching, caching and RAG.”
The ad says“Experience with cloud AI infrastructure on AWS, GCP or Azure, and with inference serving frameworks such as vLLM, TGI or TorchServe.”
The ad says“Strong software engineering fundamentals. Python proficiency is a given and familiarity with MLOps tooling such as MLflow, Weights & Biases or Ray is expected.”

About Us

Polus is pioneering the Public Safety marketplace. Our products are used by Search & Rescue, first responders, and disaster relief teams globally. We are developing radio equipment that is compatible with cellular technologies. Our goal is simple. We aim to reinvent the way Public Safety is perceived and approached. We are tireless in our pursuit of connecting communities and ensuring a safer tomorrow, today.

As our AI Optimisation Engineer, you will own and lead the AI performance agenda across our entire platform portfolio. From identifying inefficiencies to shipping optimisation strategies, your work will directly impact the technology our users depend on in the field. You will report directly to the CEO.

The Role

This is a permanent, senior position with a company-wide remit. You will be the person Polus depends on day-to-day to keep our AI running efficiently, performing well and improving over time. As our products evolve and our use of AI grows, this role grows with it. You will have a direct line to the CEO, a seat at the table with engineering and product leadership, and the authority to set direction on how we approach AI across the business. We are looking for someone who wants to build something lasting, not just tick off a project and move on.

What You Will Do

Own AI performance across the business. You are the person who knows where we stand, where things are improving and where we need to do better.
Continuously monitor, test and improve how AI runs across all our platforms, treating optimisation as a permanent function rather than a one-off exercise.
Maintain and evolve the standards, tools and benchmarks we use to measure AI performance, keeping them current as our products and models change.
Sit at the table with product and engineering leads as a standing contributor, shaping how we build and make sure AI efficiency is considered at every stage.
Hold long-term ownership of our AI cost and performance, making the day-to-day calls on the balance between speed, accuracy and cost as the business grows.
Stay ahead of developments in AI models, tools and frameworks and continuously feed that knowledge back into how we operate.
Coach and upskill engineers across teams as an ongoing responsibility, raising the bar on AI knowledge throughout the company over time.
Build a function around AI optimisation. This role grows with the company and over time you may build a team around you.

What We Are Looking For

Essential

Proven experience optimising large-scale AI/ML systems in production. You have shipped real improvements, not just prototypes.
Hands-on knowledge of LLMs, transformer architectures and the trade-offs involved in deployment, including quantisation, distillation, batching, caching and RAG.
Strong software engineering fundamentals. Python proficiency is a given and familiarity with MLOps tooling such as MLflow, Weights & Biases or Ray is expected.
Experience with cloud AI infrastructure on AWS, GCP or Azure, and with inference serving frameworks such as vLLM, TGI or TorchServe.
A structured, data-driven approach to problem-solving. You define metrics before you start optimising and you can make a clear business case for your work.
Good communication skills and the confidence to influence across engineering, product and leadership without needing a formal mandate.

What Success Looks Like

In your first 90 days you will have a solid understanding of where we stand across our AI platforms, know how we will measure progress going forward, and have started shaping the way we work. Within six months you will be a core part of how engineering and product decisions get made, with real improvements showing across at least two of our core products. Within a year, the way Polus approaches AI will look different because of the work you have done. This is a role with staying power.

What We Offer

Competitive salary and equity package.
Remote-friendly with flexible working.
A genuine seat at the table. Your decisions will shape how we build AI for years to come.
Access to the latest models, compute resources and research partnerships.
A mission that matters. Our work supports the people who keep communities safe.
Flat structure, fast decisions and no bureaucracy.

Note: This is an opportunity to work under a Swiss work agreement, we would take care of sponsoring your G-permit application (only for EU Citizens), after which you would enjoy all the benefits of working under a Swiss model.