Machine Learning Engineer - Inference / Serving

Job Description

Yobi is a rapidly growing Behavioral AI company on a mission to ethically democratize the benefits of data and AI.

Since 2019, we have built one of the largest consented behavioral datasets in the United States, extending far beyond the walled gardens of Big Tech. Unlike traditional LLM companies, Yobi builds foundation models of human behavior grounded in real-world actions such as purchases and store visits.

Our private-by-design modeling enables state-of-the-art personalization and decisioning for leading brands and agencies while protecting privacy, safety, and ethics.

Today, we are focused on bringing the performance of closed-web user acquisition to the open web and connected TV, giving brands walled-garden results without the walls.

At our core, Yobi is building the behavioral intelligence layer for any system that makes a personalization decision.

Working at Yobi

We’re at an inflection point—customer adoption is accelerating, but there’s still room to shape the architecture and culture from the ground up. Engineers here own major surface areas, build 0→1 systems in large-scale data and model infrastructure, and help define how Behavioral AI scales ethically and effectively.

Highlights:

• Partnerships with Microsoft and Databricks
• Fully remote or hybrid from several hubs (SF Bay Area, Seattle, NYC)
• World-class team of Machine Learning experts who worked on cutting edge infra and recommender systems @ Amazon, Uber, Twitter, Meta, etc.
• Product and Go-To-Market teams who have taken ideas from concept to 9 figure revenue streams

Benefits:

• Competitive Base Salary
• Meaningful equity & financial upside - a real % of the company
• Annual bonus target based on personal and company performance
• Health, Dental, Vision available
• Generous parental leave policy

About The Role

As a Machine Learning Engineer focused on Inference and Serving at Yobi, you’ll design, optimize, and operate the systems that bring our Behavioral AI models to life in real time. You’ll work at the core of our production environment, turning trained models into performant, reliable, and continuously improving services that power our open-web and CTV products.

This is an applied ML systems role—equal parts engineering depth, deployment craft, and model intuition. You’ll shape how models are packaged, versioned, rolled out, and observed across environments, ensuring every prediction is fast, accurate, and accountable.

What it takes to succeed in this role:

A reasonable estimate of the current base salary range at the time of posting is below. Base salary does not include other forms of compensation or benefits. Actual base salary within the specified range is comprised of several components, including but not limited to applicant's skill, prior relevant experience, specific degrees and certifications, job responsibilities, market considerations and the location of the position.

Base salary range: $180,000-$275,000

We prioritize attitude, culture, and general (technical) fit over matching perfectly into one of our job descriptions. If our mission and work resonates with you, we encourage you to apply. Tell us how you can help drive our products forward, even if you don’t feel like you are a perfect fit for some of the listings.

• Deep expertise in model deployment. You’ve built or scaled production ML serving systems—handling versioning, rollouts, rollback strategies, and live experimentation.
• Low-latency mindset. You understand what makes inference fast: model graph optimization, quantization, caching, batching, and efficient feature retrieval.
• Systems fluency. You write robust, high-performance code in Go, Rust, C++, or Java, and are comfortable bridging to Python for model integration and analysis.
• Operational maturity. You treat inference as a living system—monitoring drift, tracking model lineage, and ensuring observability from input to outcome.
• Infrastructure intuition. You know how to make serving systems reproducible and portable without over-engineering them, whether that’s through custom runtime design, model registries, or lightweight orchestration.
• Applied ML understanding. You can reason about model performance, interpret trade-offs, and work with researchers to make models more deployable.