Principal AI Engineer

Catapult is building the future of sports performance technology, with a mission to Unleash the Potential of every athlete and team on earth. We don't just work in the sporting industry; we are actively changing it.  Since 2006, our solutions have been leading the way in sports performance software, science, and data, in a world where 1% can literally mean the difference between winning and losing.

We work with over 5,000+ teams around the world, empowering coaches, managers and trainers in premier teams in the NFL, NBA, NHL, MLS, EPL, AFL, NRL, NCAA and more. We provide the information they need to optimize athletes’ health, game-day readiness, and performance, as well as in-game tactics.  

Catapult is a sports technology company that empowers professional teams to make data-driven decisions. We deliver health, performance, video, and AI insights from the locker room to competitive environments, ensuring every decision is an opportunity to gain an advantage, sharpen performance, and build lasting success. 

We are looking for a Principal AI Engineer who thinks in graphs, builds in Go, and believes the value of an AI system lives in the knowledge structure underneath it, not the language model on top. Based in Melbourne, you will own the core intelligence architecture for Catapult's next-generation platform: connecting the measurements we're known for into a system that reasons about sport. This is a small team with executive sponsorship, building at the intersection of knowledge graphs, LLM reasoning, and elite sports data. If you care more about getting the architecture right than following a process, we should talk.

We are building a system that reasons about sport. Not a chatbot on top of dashboards. Not a retrieval system that finds the right spreadsheet. A sensemaking system that connects all the measurements Catapult is known for into a knowledge graph, and lets anyone from the GM to the sideline coach ask questions in the language of their sport and get evidence-backed answers.

The major cloud and enterprise AI companies have validated this architecture pattern in production. Nobody has built it for sport. We are.

You will own the core intelligence architecture: the Neo4j knowledge graph, the context synthesis pipeline (map-reduce over graph communities), the agent orchestration, and the evaluation framework that keeps it honest. You are the most senior technical person on a small team. There is no technical hierarchy above you. The Head of AI sets direction, not architecture. You make the technical calls.

The stack: Go for the intelligence service, Neo4j for the knowledge graph, AWS Bedrock for model hosting, PromptFoo for eval, Python for data pipelines, React for practitioner tools, Rust for performance-critical paths. You don't need to know all of these on day one. You need to be able to learn any of them in a week.

• →Design and build the knowledge graph architecture in Neo4j. A structured representation of how sports data, practitioner behaviour, and domain concepts relate to each other. Temporal facts, community detection, confidence scoring.
• →Build the context synthesis pipeline: the system that assembles the right subgraph into the right context for the right query. Map-reduce over graph communities. The quality of the agent's output is limited by the quality of the context you assemble.
• →Build the intelligence service in Go that turns structured knowledge into practitioner-facing answers. Not text summaries. Rich, evidence-backed responses with provenance, a practitioner can click through.
• →Build the integration layer that connects Catapult's product data (wearables, gym, video, positioning) into the graph. Multiple sources, different frequencies, different schemas, one connected knowledge layer.
• →Build evaluation infrastructure from day one. Eval-driven development: define what good looks like before building, iterate until evals pass, monitor with evals in production. PromptFoo for prompt regression. LLM-as-judge for sensemaking quality. Per-step reliability measurement across the pipeline.
• →Build internal validation tools so domain experts can see, challenge, and correct the system's output. The correction IS the product's most valuable training signal.
• →Set the technical standard for a team that uses AI coding tools as the primary development environment, not a supplement. Multiple parallel sessions. Written communication over meetings. Async-first.
• →Sit on the architecture committee. You connect this team to the broader engineering organisation's technical governance.

• You have built systems that use LLMs for reasoning over structured data. At work, on a side project, or at 2am because you couldn't stop. Not chatbots. Not RAG over documents. Systems where an LLM reasons over connected knowledge and produces answers from multiple sources.
• You think in data relationships, not just tables. Graph experience is ideal but not required. Strong opinions about how to model connected data, and the curiosity to go deep on graph architectures once you're here.
• You can hold a full system architecture in your head, from data ingestion through graph construction through context assembly to user-facing response, and make trade-off decisions without waiting for approval.
• You have used AI coding tools seriously enough to have developed your own workflow. Not tried them. Developed a workflow. You have opinions about eval-driven development.
• You are comfortable being the most senior technical person in a small team with no tech lead above you.
• Strong fundamentals in at least one systems language (Go, Rust, C++) and one scripting language (Python, TypeScript). We care about engineering judgment more than language fluency.

• Experience with Neo4j or other graph databases in production.
• Experience with GraphRAG patterns, community detection, or knowledge graph construction from unstructured data.
• Experience with real-time data pipelines and temporal alignment of multi-source, multi-frequency data.
• Experience building LLM evaluation pipelines (PromptFoo, custom eval frameworks, automated scoring, regression detection).
• Experience with AWS Bedrock, model hosting, or multi-model orchestration.
• Familiarity with sports data, sensor systems, or wearable technology. Not required, but helpful.

• You need a tech lead to review your architecture decisions.
• You have only worked with LLMs through wrapper libraries and haven't thought about prompt engineering, evaluation, or cost optimisation.
• You are looking for a research role. This is applied engineering; shipping products.
• You prefer working in a well-defined system with clear tickets and sprint ceremonies. We're running like an R&D team, not a feature shipping team. Our practices are built on hypothesis testing and experiments, not sprints and story points.

• We have amazing people. We promise you’ll work with some of the most ambitious,  intelligent people in an exciting industry, and do some of the best work of your life.
• We encourage our people to engage in constructive, open, and honest communication to make Catapult extraordinary. 
• We work in a collaborative yet challenging environment to consistently improve our performance, which in turn impacts our customers' performance.
• Our workforce spans more than 20 countries. You'll have the opportunity to work with people from multiple nationalities and cultures, and to build your global awareness.  
• We value improvement and development. We are challenging ourselves to continuously grow and become a high-performance company. That means we maintain a growth mindset in everything we do, and our people are always looking for ways to improve.  There is an unlimited opportunity to grow, do more, and do better.

Whether you’re interested in sports or not, you’ll have the satisfaction of knowing your work is supporting some of the most successful teams and athletes on the planet! 

Research shows that while men apply for jobs when they meet an average of 60% of the criteria, women and other marginalized groups tend only to apply when they check every box. So if you have what it takes, but don't meet every single point in our job ad, please still get in touch! We would love to have a chat and see if you could be a great addition to our team. We are building the future of sports performance. Our priority is to find the brightest talent who can add to our team culture, actively contribute, and be excited about what they do.

All offers of employment are subject to Catapult's positive prehire check. To find out more, please contact the Talent Partner for this role.