ML Encoder

 
ML Encoder Lead Senior Machine Learning Scientist | Representation Learning & Customer Embeddings
Location: Bay Area, CA (Hybrid: Minimum 3 days per week onsite)
About the Role
We are seeking a highly experienced ML Encoder Lead to design and build a next-generation customer representation platform that powers AI, analytics, and personalization initiatives across the organization.

This role will focus on creating a shared learned representation of customers through advanced embedding and encoder models trained on longitudinal transaction, sales, engagement, and interaction data. The objective is to generate a reusable customer embedding that enables downstream machine learning, analytics, and Generative AI applications to leverage a unified understanding of customer behavior.

This is a hands-on senior-level contract position requiring both strategic and technical leadership. You will define modeling objectives, design pretraining approaches, establish evaluation frameworks, develop production-ready code, and determine through rigorous experimentation whether the approach should scale. Success will be measured not only by model quality, but by the strength and credibility of the evaluation methodology.
Key Responsibilities

• Design and implement customer encoder architectures and embedding models for large-scale behavioral data.
• Define self-supervised, contrastive, or representation learning objectives for model pretraining.
• Develop reusable customer representations based on transaction history, sales activity, engagement records, and interaction data.
• Conduct rigorous experimentation to assess whether embeddings provide measurable value to downstream AI and analytics applications.
• Establish comprehensive evaluation frameworks, including:
  • Time-based validation
  • Leakage detection
  • Cold-start testing
  • Transfer learning assessments
  • Held-out population evaluation
  • Uncertainty quantification
• Measure embedding effectiveness through calibration, stability, drift analysis, subgroup performance, and downstream model impact.
• Build scalable training pipelines and production-ready machine learning systems.
• Partner with business and technical stakeholders to communicate findings, risks, and recommendations.
• Provide objective guidance, including recommending discontinuation of approaches that do not demonstrate sufficient value.

Required Qualifications

• Proven experience personally designing and training encoder or embedding models, including creation of pretraining objectives.
• Deep expertise in:
  • Representation Learning
  • Self-Supervised Learning
  • Contrastive Learning
  • Transformer Architectures
  • Temporal and Sequential Modeling
  • Graph Neural Networks (GNNs)
  • Recommender Systems
• Experience working with large-scale, sparse, longitudinal event data, such as:
  • Customer journeys
  • Transactions
  • Claims
  • Clickstream data
  • Engagement histories
• Experience building inductive representations capable of modeling entities with limited historical data.
• Strong understanding of experimental design, model validation, and machine learning evaluation best practices.
• Ability to assess true incremental business value of learned representations across downstream applications.
• Advanced Python development skills and hands-on experience with:
  • PyTorch and/or JAX
  • SQL
  • Distributed data processing
  • Cloud-based machine learning platforms
• Experience deploying models from research through production, including:
  • Data contracts
  • Training pipelines
  • Model versioning
  • Serving infrastructure
  • Monitoring
  • Reproducibility frameworks
• Strong communication skills with the ability to present complex findings, limitations, and uncertainty to executive and technical audiences.

Preferred Qualifications

• Experience developing Customer 360 solutions, behavioral embeddings, recommender systems, or foundation models trained on event-based data.
• Knowledge of privacy, fairness, bias mitigation, and re-identification risk in customer representation models.
• Publications, patents, open-source contributions, or publicly recognized work in machine learning and representation learning.
• Experience in industries with large-scale behavioral data, including:
  • Consumer Technology
  • Marketplaces
  • Streaming Platforms
  • Financial Services
  • Payments
  • Advertising Technology

Industry-specific experience is helpful but not required. Candidates with strong representation learning expertise from any high-scale data environment are encouraged to apply.
Keywords
Machine Learning Scientist, ML Engineer, Representation Learning, Embeddings, Encoder Models, Self-Supervised Learning, Contrastive Learning, Customer 360, Recommender Systems, Graph Neural Networks, Transformers, PyTorch, JAX, Behavioral Modeling, Customer Analytics, Generative AI, Foundation Models, MLOps, Distributed Training, Machine Learning Research, AI Engineering, Data Science, Cloud ML, Feature Learning, Production Machine Learning
Why This Opportunity?
This role offers the chance to define the foundational customer intelligence layer that powers future AI and analytics products. You'll lead cutting-edge representation learning initiatives, influence technical strategy, and help establish whether a transformative machine learning capability becomes a core organizational asset.