Staff Machine Learning Engineer

CloudX is building an AI-native supply-side advertising platform for mobile publishers. The Staff Machine Learning Engineer will be the first dedicated ML hire, owning the design, deployment, and optimization of machine learning systems that maximize publisher revenue and customer lifetime value. The role also includes product ownership, technical leadership, and mentoring future ML engineers.

Responsibilities:

• Machine Learning Engineering: design, train, evaluate, and ship the models that power the revenue-optimization product. You'll own the full lifecycle, from feature definition through production deployment and online evaluation. You'll make the architectural calls — what models, what training framework, what serving approach — and you'll write the code to make them real
• Product Ownership: the "make me more money" button is a multi-year product surface, starting with floor pricing, extending into waterfall and bidder-order optimization, and eventually joint optimization across the full set of publisher controls. You'll work directly with Product and with publishers to understand what's actually worth optimizing for and sequence the roadmap accordingly
• Technical Leadership: lead by example to build out the ML discipline at CloudX. Today, several engineers across backend and infra contribute to the ML effort as part of their broader work; you'll be the person setting direction, raising the bar, and — as the function grows — helping us hire and mentor additional ML engineers

Requirements:

• We're looking for someone who has hands-on experience actually training and deploying traditional machine-learning models
• You've shipped ML into a production request path. Not a batch job, not a notebook, not a dashboard. A model serving real traffic under a latency SLO, where getting it wrong costs money. You can talk about a specific system you built, the lift you measured, and how you measured it
• You've owned the offline-to-online feature parity problem. You've seen training/serving skew, you've written (or reviewed) the featurizer that runs in both places, and you have a view on how to keep them consistent as the system evolves
• You've run real experiments measuring real revenue impact. You understand the difference between "the model log-likelihood improved" and "the business made more money," and you can describe a time those disagreed and what you did about it
• You can get comfortable outside Python. You tell us what you need for the models, but the rest of our services are mostly Golang. You don't need to be a Go engineer, but you should be willing to learn enough to read production serving code, flag where it diverges from training, and contribute fixes when it does
• Hands-on expertise: in case we weren't clear enough already, this is a hands-on position. You may have managed or led ML teams at points in your career, but you still code regularly and are interested in continuing to do so. You've owned large projects end-to-end and know how to work well with others
• Strong written communication skills: you are used to writing about, speaking about, and generally communicating complex technical subject matter both to other engineers and to non-engineers
• Early-stage mentality: you understand that success at a startup involves grit and determination. You have good taste when it comes to trading off speed vs. perfection. You know when to cut corners but aren't afraid to advocate for rigor when you believe it's necessary
• AI forward: you are actively experimenting with or using AI as part of your software engineering practice. You don't send vibe-coded slop to your teammates to review, but you use AI appropriately to achieve great results
• High ownership: you care a lot about your work and when you ship a product, you make sure it continues to solve problems for the customer. You care a lot about the customer, the overall business, and are constantly trying to help achieve success — with or without code
• If you're not sure if this role is an exact fit, we encourage you to apply. Many members of our team have had interesting career paths and we relish the chance to work with extraordinary individuals
• Adtech experience: you've worked in adtech — SSP, DSP, ad exchange, RTB — and have a good understanding of the broader ecosystem and market. Equivalent experience from other low-latency, revenue-objective ML domains (search ranking, recsys, marketplace pricing for rides/delivery/lodging, or quant execution) is a real substitute and we'll treat it as such
• Auction and bidding model experience: hands-on experience with contextual bandits, reinforcement learning, Thompson sampling, or other approaches that fit the explore/exploit structure of auction pricing. Familiarity with the RTB literature or systems like Meta's Pearl is a strong positive
• Stack experience: we're running in AWS, our inference is ONNX-based, we currently train with XGBoost (and are evaluating LightGBM), we use ClickHouse for analytics, Kubernetes for training orchestration, and Datadog for observability. All of this is v0; we'd be happy to speak with you if you have strong opinions about the right tools for the job