ML Encoder Lead (Senior Role)

This position will report to a client located in South San Francisco, CA. Local candidates able to have an onsite presence as needed are preferred. Well-qualified remote candidates able to overlap with PST working hours are also considered.

Pay

$100–$200/hour W2, based on qualifications.

Schedule

6-month contract to start; full-time hours overlapping PST business hours.

Responsibilities

• Design and develop customer representation-learning models using longitudinal transaction, sales, and interaction data.
• Define pretraining objectives and build dense customer embeddings for downstream GenAI and analytics applications.
• Develop, train, and evaluate encoder models using large-scale sparse event data.
• Design rigorous model evaluation using time-based splits, leakage detection, cold-start analysis, held-out populations, uncertainty, and baseline comparisons.
• Assess embedding quality through downstream signal, calibration, stability, drift, and subgroup performance.
• Build production-ready training and serving pipelines with reproducibility, model versioning, monitoring, and data contracts.
• Lead the transition of representation-learning research into scalable production systems.
• Present modeling results and evaluation evidence to senior stakeholders and make recommendations on model continuation or termination.

Requirements

• Bachelor’s or Master’s degree in Computer Science, Machine Learning, Artificial Intelligence, Statistics, or a related technical field.
• Hands-on experience personally training encoder or embedding models and designing pretraining objectives.
• Deep expertise in representation learning, including self-supervised or contrastive learning.
• Strong experience with sequence/temporal modeling, Transformers, Graph Neural Networks, or recommender-system embeddings.
• Experience modeling large-scale, sparse, longitudinal event data such as transactions, claims, clickstream, customer journeys, or engagement histories.
• Experience developing inductive representations for cold-start entities using entity features rather than lookup tables.
• Advanced Python with PyTorch or JAX.
• Strong SQL and distributed data processing experience.
• Experience with cloud-based model training at scale.
• Experience building production ML pipelines, model serving, versioning, monitoring, and reproducible training workflows.

Preferred Skills

• Customer-360 representations and behavioral embeddings.
• Recommender systems or foundation models applied to event data.
• Experience addressing privacy, fairness, and re-identification risks in learned representations.
• Publications, patents, or public applied work in representation learning.
• Experience with large-scale behavioral data in consumer technology, marketplaces, streaming, financial services, payments, or advertising technology.