Data Scientist, Applied AI & Machine Learning

OEC builds the software that dealers, repair shops, suppliers, and automakers use to find, price, order, and stock auto parts. Applied AI/ML is OEC's AI & Machine Learning team. We work with product teams across the business to frame a problem, build the data science and ML behind it, ship it to production, and prove the value with real measurement. We work horizontally across the organization, with teams in Mechanical, Supply Chain, Collision, and International, primarily supporting in-product use cases. You'll join a small team of applied scientists and ML engineers across Europe and the US, and you'll report to the head of Applied AI/ML.
This is a mid-level individual contributor role with end-to-end ownership. Success in this role means you improve business metrics across our product lines and make a real, meaningful impact on the organization. Our projects cover a wide range of problems, so we'll match your first projects to your background. You don't need experience in every area below. If you meet most of the requirements and the work interests you, please apply.
What You'll Do

Work on the kinds of problems we take on, depending on your background:
Search and ranking: semantic search, learning to rank, and personalized ranking to help shops find the right part faster.
Classification and matching: map parts to industry-standard part types and match attributes across catalogs.
Forecasting: forecast parts demand so dealers stock the right parts.
Customer and partner analytics: predict churn and early-tenure dropout, segment customers, and learn which suppliers and parts perform best.
Anomaly detection: find fraud, bots, and suspicious usage.
LLMs and agents: use LLMs when they're the right tool, with evaluation that measures output quality against a clear rubric.
A/B testing: help teams design and analyze A/B tests with enough statistical power to detect real effects, so we can measure and improve our products.
Recommendation engines: build recommendation services for partners, customers, and internal teams.
Work with a product team early to understand the problem, check the data, and decide whether AI/ML is a good fit for the problem statement.
Work on your models in production, including feature pipelines, training pipelines, deployment, drift monitoring, and retraining.
Measure impact with techniques such as offline evaluation, shadow mode, A/B tests, and causal methods before a change goes live.
Present results and tradeoffs to product and business stakeholders in terms they use, such as conversion, GMV, retention, and fill rate.
Write documentation and teach the product team how to support what we built, so they can own it after handoff.
Work with the Applied AI/ML team to improve best practices, evaluate new tools, and bring new ideas to the table.

What You Bring

At least 3 years of applied data science or machine learning experience.
At least one model that you took to production and that moved a measurable business metric.
Strong Python and SQL, with production-quality code, testing habits, and version control.
A solid base in statistics and machine learning, and good judgment about when a simple approach is enough.
Hands-on experience in at least one of the problem areas above, such as ranking, recommendation, classification, forecasting, or churn modeling.
Experience building data pipelines and deploying models on a cloud ML platform, such as Amazon SageMaker, Azure ML, or Databricks. We use AWS (including SageMaker), Airflow, and MLflow. Equivalent tools also count as experience.
Comfort with large, messy transactional and behavioral data.
Clear written and spoken English.
Experience explaining tradeoffs to non-technical stakeholders.

Nice to Have

Master's degree or PhD in a quantitative field.
Search relevance or recommender systems in e-commerce or a marketplace, including learning to rank, embedding-based retrieval, or candidate generation with re-ranking.
Power analysis and causal inference methods.
LLMs in production and evaluation harnesses for LLM output.
Real-time model serving and low-latency inference, in addition to batch scoring.
Experience with experiment design and offline evaluation.
Amazon SageMaker
Terraform, infrastructure as code.
Retail media, sponsored placement, or advertising relevance.
Automotive, automotive e-commerce, or B2B parts and supply chain, including complex product catalogs and taxonomies. Knowledge of industry standards such as ACES, PIES, or PCDB, or of OE catalogs.