Applied Scientist; Perm

Position: Staff Applied Scientist (Perm or Contract)

Location: Greater London

About Us

Permutive’s mission is to rebuild data in advertising to protect privacy. The open internet needs a sustainable foundation that funds the free & open online experiences we all rely on.

Permutive’s data collaboration platform uses edge technology to securely process data where it is and AI to discover signals that drive performance—respecting consumers’ privacy preferences while delivering outcomes for publishers & advertisers. Our customers include some of the world’s largest media companies and advertisers—including News Corp, Warner Brothers Discovery, Hearst, The Guardian, Sky, State Street, Sonos, and Apple—and we’re backed by leading investors like Soft Bank Vision Fund and Y Combinator.

Engineering

At Permutive

We’re 30+ engineers working to build an outstanding engineering culture so that everyone who joins has the opportunity and the support to do the best work of their life. Small, autonomous teams are important to us, and we want to empower everyone to make decisions confidently and take ownership of their impact.

We operate at a massive scale: our platform handles more requests each day than there are new tweets and Google searches, and each month we serve more than 2 billion user devices. We combine our patented edge‑computing capabilities and low‑latency cloud services to deliver real‑time customer experiences.

We think applying functional programming techniques like compositionality and type‑safety is the best way to build the type of massively distributed system our platform comprises, allowing us to move fast without sacrificing quality.

About

The Role

We're looking for a Staff Applied Scientist to own optimisation across our platform. You'll be the domain authority on how we use data to make campaigns perform better — from how audiences are composed and scored, to how inventory is priced and allocated, to how campaigns are tuned over their lifetime.

The scope is broad: campaign optimisation, audience quality modelling, lookalike improvement, and measurement all fall within your remit. You'll work across teams, identifying which problems to solve, designing the approaches, and owning the models through to production.

This is a hands‑on technical leadership role with broad scope and real autonomy. You'll operate across the full platform, with the freedom to identify the highest‑leverage problems and design the approaches to solve them. You'll shape what we build and have a seat at the table when we make product and strategy decisions.

What you’ll be doing ?

• Design and build models for campaign optimisation, working across metrics like CTR, CPA, and ROAS using publisher‑side signals (cohort membership, contextual data, engagement) combined with advertiser outcome data.
• Develop approaches to inventory allocation that account for audience quality, placement value, and supply dynamics.
• Own your models end‑to‑end from conception through production monitoring. You'll work closely with specialist engineers who integrate models into our high‑throughput Scala services, SDKs, and data pipelines — you own the logic, they own the infrastructure, with shared accountability for outcomes.
• Influence product direction by identifying opportunities, framing feasibility, and bringing evidence to roadmap decisions.
• Run experiments and prototypes to validate ideas quickly, then evolve the best ones into production systems.

What you’ll need? Essential

• Direct experience with optimisation or bidding systems in programmatic advertising — within a DSP, SSP, ad server, or comparable platform. You understand the mechanics of how campaigns are priced, paced, and optimised in practice.
• Strong applied ML and statistical foundations. You can design, evaluate, and iterate on models — and you know when a heuristic beats a neural network.
• Production ML experience. You've built models that run in real systems, not just notebooks. You understand latency, data freshness, feedback loops, and the gap between offline evaluation and online performance.
• Fluency with the programmatic ecosystem — bid requests, win notifications, the signals available at decision time, and the differences between publisher‑side and buy‑side data.
• St…