Staff Applied Scientist (Perm or Contract)

Overview

As Staff ML Engineer at Permutive, you will own platform-wide optimisation by building and refining models that improve campaign performance. You’ll shape how audiences are composed, inventory is allocated, and campaigns learn over time, influencing product and strategy. You’ll work across teams with autonomy, delivering end-to-end models from conception to production. This role offers the chance to tackle novel modelling challenges in a privacy-first, edge-enabled platform at scale.

Pay / Benefits

• Stock options
• Fully paid parental leave
• Flexible hours
• Home office budget
• Unlimited paid time off with minimum 25 days
• Ongoing training and development opportunities

Responsibilities

• Design and build models for campaign optimisation using CTR, CPA, and ROAS metrics with publisher signals and advertiser data
• Develop inventory allocation approaches considering audience quality and supply dynamics
• Own models end-to-end from conception to production monitoring, collaborating with engineers to integrate into high-throughput services
• Influence product direction by framing feasibility and driving roadmap decisions with evidence
• Run experiments and prototypes to validate ideas and evolve successful approaches into production systems

Key requirements

• Direct experience with optimisation or bidding systems in programmatic advertising (DSP/SSP/ad server)
• Strong applied ML and statistical foundations; ability to design, evaluate, and iterate models
• Production ML experience; systems-level understanding of latency, data freshness, feedback loops
• Fluency with programmatic ecosystem; decision-time signals, bid requests, win notifications
• Strong communication skills for explaining concepts to engineers, PMs, and leadership

• Strong communication
• Cross-team collaboration
• Problem-solving mindset
• Applied machine learning and statistical modelling
• Experience with production ML systems and latency considerations
• Optimization or bidding systems in programmatic advertising

…