Position Title: ML Encoder Lead (Senior Role) Work Location: San Francisco, CA Bay Area Assignment Duration: 12 Months Work Arrangement: Hybrid – Bay Area-based, with ability to work from the office minimum of 3 days per week Position Summary: The goal is to build the first shared learned representation of our customers — one dense vector per customer, trained on longitudinal transaction, sales and interaction history — that downstream GenAI and analytics products can reuse instead of each re-deriving its own view of the same market. The contractor will design the pretraining objective, train and evaluate the encoder, and produce the evidence that determines whether the approach continues. Evaluation is as much of the deliverable as the model. This is a hands-on senior contractor who must define the modeling objectives and evaluation design and write production code — not execute a specification handed to them. Key Responsibilities: Design the pretraining objective, train and evaluate the encoder, and produce the evidence that determines whether the approach continues. Define the modeling objectives and evaluation design and write production code. Qualification & Experience: Minimum capabilities Has personally trained an encoder or embedding model, including designing the pretraining objective — not only consumed pre-trained embeddings or fine-tuned a published large language model Deep expertise in representation learning: self-supervised or contrastive pretraining, sequence and temporal modeling, transformers, graph neural networks or recommender embeddings Experience modeling large, sparse, longitudinal event data such as transactions, claims, clickstream, customer journeys or engagement histories Experience building inductive representations, so an entity with little history can be represented from its own features rather than a lookup table Rigorous evaluation practice: time-based splits, leakage detection, cold-start slices, transfer to held-out populations, stated uncertainty and hard baselines Ability to judge whether an embedding carries genuine incremental signal downstream, including calibration, stability, drift and subgroup performance Strong Python engineering with PyTorch or JAX, SQL, distributed data processing and cloud-based model training at scale Experience carrying a model from research into production: data contracts, training pipelines, versioning, serving, monitoring and reproducibility Ability to present findings and uncertainty credibly to senior stakeholders, and to recommend stopping an approach that is not working Preferred Experience with customer-360 representations, behavioral embeddings, recommender systems or foundation models over event data Familiarity with privacy, fairness and re-identification risk in learned representations of individuals Publications, patents or public applied work in representation learning Any industry with large-scale behavioral event data is relevant — consumer technology, marketplaces, streaming, financial services, payments or advertising technology. Domain knowledge is not required.