Applied Machine Learning Scientist (Across Various Levels)

Role overview

This opportunity is for applied machine learning scientists from associate through principal/lead level, with the title and scope adjusted to fit the candidate’s background. The team is focused on taking machine learning models into real production environments, where performance, reliability, and latency all matter at scale.

The role is part of a product-focused team that builds and operates ML systems end to end. Rather than research prototypes or short-lived proofs of concept, the work here goes directly into live products used by millions of people. If you want to build models that are deployed, monitored, and iterated in the real world, this role is designed for that.

What you will work on

You will create, train, refine, and ship machine learning models across areas such as recommendation, personalization, computer vision, natural language processing, and forecasting. These systems are expected to reach web, mobile, and broadcast experiences within weeks and support more than 10 million users.

The problems are practical and complex: cold-start scenarios, heavily imbalanced datasets, strict 100 ms latency requirements, regulatory limitations, and editorial policy constraints. You will work closely with Product and Engineering, participate directly in business discussions, and join weekly reviews with the CTO. The expectation is that you do more than hand off a model — you help instrument it, deploy it, and take ownership of its outcomes.

Tech stack and environment

The team works with a modern ML and MLOps stack that includes PyTorch, TensorFlow, JAX, Transformers, and LLMs, as well as recommendation systems, computer vision, and graph neural networks depending on the team’s needs.

Operational tooling includes Kubernetes, Ray, Kubeflow, MLflow, Feast, Triton, and vLLM. Data infrastructure spans Spark, Kafka, BigQuery or Snowflake, and vector databases, all running on GCP or AWS at real-time scale. You do not need to know every tool on day one, but you should be able to learn quickly and go deep where needed.

What success looks like

The team values strong growth and practical execution. You are likely to do well if you have already delivered ML systems into production and have experience with model drift, experimentation, rollback strategies, and situations where a newer model performs worse than the previous version.

Strong candidates think in trade-offs, can explain why one model approach outperformed another, and write production-grade Python with tests. They understand latency budgets such as p99 under 200 ms and how those requirements influence model architecture and serving design. A product-first mindset is important, including the ability to ask whether something should be built before focusing on whether it can be built.

Continuous learning is also essential, since the ML landscape and the roadmap both evolve quickly. The team hires for growth potential as well as current expertise.

Background and qualifications

A Master’s or PhD in Machine Learning, Computer Science, Statistics, or Mathematics is helpful, but strong equivalent practical experience is also accepted. Demonstrated work on impactful ML systems matters just as much as formal credentials.

Experience in media, recommendation systems, advertising, or robotics is a plus, though it is not required. Publications, competitive Kaggle results, or substantial personal/open-source projects are also considered valuable evidence of skill and initiative.

Why this role stands out

This is a chance to build models that influence millions of users rather than small-scale experiments. Successful work is visible to senior leadership, and the team culture emphasizes rigorous code review and strong technical standards.

The work includes challenges such as cold-start recommendations for new content, multi-objective recommender systems, live AI moderation, and controlling inference cost at broadcast scale. The role also offers room to grow into Principal Scientist, transition into Engineering Management, or move toward product-focused leadership.

Compensation is positioned to reflect the level of expertise involved, with a competitive base salary, equity, and bonus. Additional support includes access to GPUs on demand and a budget for conferences. The team does not rely on separate “innovation days” because shipping is considered the core innovation.

Application instructions

Applicants should submit a CV along with two short bullets describing the most impactful ML system they have delivered, plus one problem they would most like to solve for the client.

Equal opportunity

The employer follows an equal opportunity policy and is committed to maintaining an inclusive workplace for everyone.

Notification

Only candidates selected for the next stage will be contacted.