AI Product Manager (Technical) Perth, WA · Full-Time

About the Company

Our client uses AI to keep people safe across large-scale surveillance networks. Coming off their strongest financial year, they're building something important and large out of Perth. They're looking for a Technical AI Product Manager to maintain their lead as their product suite and customer base expands.

The Role

You'll drive the definition and delivery of AI/analytics capabilities and the product experiences around them, working closely with the existing Product Manager (overall strategy/roadmap) and Engineering Manager (execution and delivery). Your focus is turning customer and operational needs into technically sound AI product requirements - and ensuring what ships is evaluated, monitored, and improved in production.

This is a product role with serious technical depth. You won't be writing models day-to-day, but you'll be comfortable shaping how models are evaluated, deployed, and monitored, and how users build trust in the outputs.

You genuinely love AI. Not just professionally — you're the kind of person who has Claude, ChatGPT, and a half-built LLM side project open on your laptop on a Sunday. You bring that curiosity and hands-on instinct into every product decision.

What You'll Do

• Own AI feature definition end-to-end: problem framing, success metrics, data requirements, evaluation plans, rollout, and iteration.
• Define evaluation strategy and acceptance criteria across offline metrics (precision/recall, ROC/PR curves, calibration, latency, robustness), operational metrics (false alarm rate, missed event rate, time-to-triage, operator adoption), and dataset strategy (gold sets, edge-case sets, drift detection sets).
• Drive data-centric product improvements: data quality requirements, labelling workflows, feedback loops, and mechanisms to capture "ground truth" from users.
• Partner with engineering and ML engineers on architecture-level trade-offs: model packaging, inference placement (edge vs server), throughput/cost constraints, and deployment cadence.
• Own AI observability in production: monitoring for drift, performance degradation, data pipeline failures, and model/version tracking. Define alerting and dashboards.
• Contribute to MLOps and release processes: model versioning, rollout/rollback, canarying, guardrails, documentation, and customer communication.
• Work with customers and internal teams to validate outcomes in real deployments - different sites, lighting, camera angles, behaviours, and operational constraints.

What You'll Bring

• 3+ years shipping B2B software as a Product Manager (or similar), with meaningful AI/ML/analytics exposure.
• Strong understanding of ML product concepts: classification vs anomaly detection, thresholds and trade-offs, precision/recall, class imbalance, base rates, dataset bias, drift, data quality constraints, and evaluation design pitfalls.
• Ability to write technical product requirements that include data inputs/assumptions, interfaces, performance targets, evaluation methodology, "definition of done", failure modes and guardrails.
• Comfort working in engineering tools and artefacts (Jira/Linear, API docs, logs, dashboards) and discussing systems constraints (latency, scale, cost).
• Excellent communication skills — you can align stakeholders and explain technical trade-offs clearly to non-technical audiences.

Nice to Have

• Experience with computer vision / video analytics, real-time inference systems, or security / industrial / OT domains.
• Familiarity with edge deployment, GPU/CPU inference optimisation, or streaming architectures.
• Experience with human-in-the-loop systems (review / label / feedback workflows).
• Exposure to compliance / privacy considerations in video environments.

What Success Looks Like (First 6–12 Months)

Clear AI feature roadmap inputs and PRDs with measurable metrics and evaluation plans. Improved event quality and operator trust via better tuning, explainability, and feedback loops. Production monitoring and model/version visibility that reduces "silent failures." Faster iteration cycles from customer signal to data/eval to release.

Why This Company

Real-world AI — ship products customers rely on, not just demos. High ownership - influence product direction end-to-end. Small team, low bureaucracy - move fast, collaborate closely, see impact quickly.