Senior ML Encoder Lead: Representation Learning, Production

Dawar Consulting, Inc.; South San Francisco (CA); On-site; USD 140,000 - 190,000; Full time

Benefits offered by this job

Medical

Paid Sick Leave

401 (k)

Job summary

Dawar Consulting, Inc. in South San Francisco, CA is seeking a Senior ML Encoder Lead to develop shared customer representations from longitudinal data, define objectives, and build embedding models.

You will produce production-ready code, evaluate models with time-based splits, and present findings to senior stakeholders. The role emphasizes scalable ML pipelines, deployment readiness, and collaboration with cross-functional teams.

Qualifications

• Proven experience training encoder/embedding models and pretraining objectives.
• Expertise in representation learning (self-supervised/transformers).
• Experience with large-scale, sparse longitudinal data.
• Ability to develop inductive representations with limited history.
• Strong evaluation: time-based splits, leakage detection, robust baselines.
• Strong Python skills with PyTorch or JAX, SQL, cloud training.
• Experience bringing ML from research to production with pipelines and monitoring.

Responsibilities

• Define modeling objectives and build encoder/embedding models.
• Develop production-ready code and data pipelines.
• Evaluate approach for downstream value and communicate results.

Skills

Encoder training

Representation learning

Time-based splits

Python

PyTorch

JAX

SQL

Distributed processing

Cloud training

Production pipelines

Communication

Tools

PyTorch

JAX

AWS

GCP

Job description

Dawar Consulting, Inc. in South San Francisco, CA is seeking a Senior ML Encoder Lead to develop shared customer representations from longitudinal data, define objectives, and build embedding models.

You will produce production-ready code, evaluate models with time-based splits, and present findings to senior stakeholders. The role emphasizes scalable ML pipelines, deployment readiness, and collaboration with cross-functional teams.