Post-Training Research Engineer: Ship Real Models

Vizcom Technologies, Inc.; San Francisco, Northern (CA, KY); Hybrid; USD 164,000 - 215,000; Full time

Benefits offered by this job

Medical coverage for employees and 25%

Equity ownership

Flexible PTO

401(k) with employer match

Learning & Development allowance

Paid parental leave

Lunch at SF headquarters

Gym membership stipend

Job summary

Vizcom is seeking a Research Engineer, Post-Training in San Francisco, CA, in-person, full-time. You will work on models that learn from designer judgments to move from pencil to product, collaborating with engineering on the post-training stack and documenting experiments for reproducibility.

Responsibilities include training and evaluating reward models, exploring fine-tuning and reinforcement learning methods, and ensuring research results translate into production if feasible.

Qualifications

• Strong programming and software engineering skills for ML systems.
• Hands-on experience training or fine-tuning ML models.
• Experience with generative models incl. diffusion or flow models.
• Familiarity with post-training methods like supervised fine-tuning, reward modeling, distillation, or reinforcement learning.
• Experience designing experiments and evaluating model performance.
• Strong ML fundamentals and turning research ideas into working implementations.
• Interest in product-coupled research informed by real users and production signals.
• Comfort solving open technical problems.
• Collaborative mindset with documentation of results.

Responsibilities

• Execute post-training research and engineering projects from design through evaluation and implementation.
• Train and evaluate reward and preference models using design decisions.
• Explore methods including supervised fine-tuning, distillation, and reinforcement learning.
• Develop rigorous evaluations to ensure reproducibility and impact.
• Collaborate with senior researchers to translate results into production systems.
• Partner with Product and Design to validate model performance in real workflows.
• Evaluate emerging post-training techniques and prototype approaches for our stack.
• Contribute to reliable training and experimentation infrastructure.
• Document experiments, results, and learnings for future work.

Skills

Programming

ML systems

Experiment design

Collaborative research

Tools

Diffusion models

Post-training methods

Job description

Vizcom is seeking a Research Engineer, Post-Training in San Francisco, CA, in-person, full-time. You will work on models that learn from designer judgments to move from pencil to product, collaborating with engineering on the post-training stack and documenting experiments for reproducibility.

Responsibilities include training and evaluating reward models, exploring fine-tuning and reinforcement learning methods, and ensuring research results translate into production if feasible.