CTV performance advertising is still being invented. The auction dynamics, identity graph, and feedback loops are nothing like display or search. There's no playbook to copy you'd be writing it. 2. You own the full decision loop. From prediction to bid price to pacing, on a product real advertisers depend on today. Your judgment shapes the direction. You'll see the impact in the numbers within weeks. 3. The timing is right. Past "does this work?" and into "how fast can we scale this?" Early enough to shape the architecture. Late enough that you're not building on sand. ## What You'll Do
Own the pacer and bidder
Go deep on the bidder and pacer stack map it end-to-end and identify at least one fragility worth owning on day one
Propose improvements to the bidder or pacer autonomously from problem definition through to design proposal
Ship measurable improvements to CPx, pacing accuracy, or win rate within your first three months
Raise the team's ceiling
Distribute critical system knowledge: make the team faster and less dependent on any single person
Challenge existing designs, including inherited ones, and push for cleaner solutions
Deepen expertise in the theoretical foundations (probabilistic modelling, PID control theory) powering our approaches
Move at Vibe pace
Use AI tooling as a genuine force multiplier (not a crutch, not a checkbox)
Iterate in production, not in your head, we - ship, measure and adjust
What You Need
Shipped backend systems in a strongly typed language C++, Rust, Go, Scala, Java, or equivalent at production scale
Debugged distributed systems under real traffic: latency issues, consistency failures, the works
Worked inside a DSP, SSP, or bidder: you understand auction mechanics from the inside
Built or tuned pacing algorithms or PID controllers theory and implementation, not just awareness
Solid grounding in probabilistic reasoning and statistical modelling you think in distributions, not just averages
Nice to Haves
Experience in a scale-up environment you know what fast and resource-constrained actually feels like
ML systems in production: feature pipelines, model serving (not just training notebooks)
Real-time data pipelines and low-latency architectures at scale
Redis, Kafka, or equivalent under meaningful load
What We Offer
Variable pay — based on objectives you hit. No arbitrary targets. - Hybrid flexibility — We're in the heart of Paris and our team is in 3x a week! - Health insurance — Full coverage via Alan. - Meal vouchers — Via Swile. - Annual offsite — The whole team, once a year, somewhere worth the trip. - Tech Syncs — Engineering and Product meet in person at least quarterly, worldwide. ## Our Interview Process
We respect your time. Here's exactly what to expect — no surprises, no ghosting. 1. Recruiter screen — 30 min. We'll share context; you'll share yours. 2. Manager interview — Meet the person you'd work with directly. 3. Technical Coding Interview and Systems Design Interview
4. Calibration interview — A senior leader joins to ensure we're holding a consistent, high bar. 5. Offer — Fast. We don't let good decisions sit. 6. Reference checks — Two calls, handled with discretion. ## Referral Instructions
Being referred? Ask your contact to submit your application directly on your behalf.
