ML Engineer, Generative Media

CGS; San Francisco (CA); On-site; USD 200,000 - 350,000; Full time

Benefits offered by this job

Visa sponsorship

Equity

Competitive salary

Job summary

CGS, a seed-stage applied AI lab in San Francisco, is seeking an ML Engineer specializing in generative media to own the models behind our AI video creation platform. You will train, evaluate and deploy models in production, collaborating with a small, highly technical team.

You'll develop proprietary lightweight models to improve quality, speed and intelligence of generated media, take experiments through production, and contribute to product features.

Qualifications

• 3–6 years of experience in machine learning, generative AI, computer vision or a closely related field.
• Hands-on experience training or fine-tuning modern image or video generation models, including diffusion, flow matching or comparable architectures.
• Proven track record of shipping machine learning models into production products.
• Experience with training pipelines and a strong understanding of training-data strategy.
• Familiarity with generative-model evaluation, benchmarks and feedback loops.
• Experience at a recognized generative-media, computer-vision or frontier AI company or research lab.
• Willingness to work in person in San Francisco five days per week; relocation and visa sponsorship available.

Responsibilities

• Train and fine-tune modern image and video generation models using diffusion and related techniques.
• Develop proprietary models to improve quality, speed and intelligence of AI-generated media.
• Take models from training through evaluation, deployment and live product integration.
• Design and refine training-data strategies including collection, labeling, filtering and feedback loops.
• Build benchmarks, evaluation frameworks and measurement systems to monitor model performance.
• Investigate model behavior and identify opportunities to improve output quality and generation speed.
• Collaborate with a small, highly technical team to translate AI advances into product features.

Skills

Machine learning

Generative AI

Computer vision

Production deployment

Model evaluation

Job description

Experience: 3–6 years

Company stage: Seed-stage AI startup

About the Company

An ambitious, seed-stage applied AI lab in San Francisco is building the next generation of AI-powered creative tools. Founded by an experienced team that has previously built and sold companies, the company is backed by well-known investors and already has strong public traction.

Its AI video creation product puts cutting-edge generative models directly into the hands of users. The team is now looking for an ML Engineer specialising in generative media to take ownership of the models that power the product, developing proprietary capabilities that make video generation smarter, faster and more capable. With a team of fewer than ten people, this is an opportunity to have a significant impact on both the technology and the product.

The Role

As an ML Engineer, Generative Media, you will own the development, training and evaluation of the models behind the company's AI video creation platform.

You'll work hands-on with modern image and video generation techniques, taking models from experimentation and training through to production deployment. Alongside improving existing capabilities, you'll develop proprietary, lightweight models designed to make the product more intelligent and efficient.

This is a highly practical, product-focused role for someone who combines deep generative-media expertise with a track record of delivering models that work in the real world.

What You'll Do

• Train and fine-tune modern image and video generation models using techniques such as diffusion and flow matching.
• Develop proprietary models that improve the quality, speed and intelligence of AI-generated media.
• Take models from training and experimentation through evaluation, deployment and integration into a live product.
• Design and refine training-data strategies, including data collection, labelling, filtering and feedback loops.
• Build benchmarks, evaluation frameworks and measurement systems to understand and improve model performance.
• Investigate model behaviour and identify opportunities to improve output quality, consistency and generation speed.
• Work closely with a small, highly technical team to translate advances in generative AI into practical product features.

What We're Looking For

• 3–6 years of relevant experience in machine learning, generative AI, computer vision or a closely related field.
• Hands-on experience training or fine-tuning modern image or video generation models, including diffusion, flow matching or comparable architectures.
• A proven track record of shipping machine learning models into production products, rather than working exclusively on theoretical research or evaluation.
• Experience with training pipelines and a strong understanding of training-data strategy.
• Familiarity with generative-model evaluation, benchmarks and feedback loops.
• Experience at a recognised generative-media, computer-vision or frontier AI company or research lab, such as Meta, ByteDance, Google DeepMind, Runway, Pika, Luma, Stability AI or a comparable organisation.
• The ability to take ownership of a model's development, from initial training through to measurable product impact.
• Willingness to work in person in San Francisco five days per week. Relocation is welcome, and visa sponsorship is available.

What Will Make You Stand Out?

• You can point to a specific image or video model, feature or capability that you personally trained and helped ship.
• You understand how training-data choices influence model behaviour and can explain what you changed and why.
• You've built evaluation harnesses, benchmarks or feedback loops that demonstrably improved generative-model performance.
• You have experience with video generation, text-to-video, image generation or other modern generative-media systems.
• Research publications or open-source contributions that complement a strong production track record.

Why Join?

• Significant ownership: Lead the development of proprietary generative-media models in a team of fewer than ten people.
• Real product impact: Your models will power a live product with strong public traction.
• Competitive compensation: Earn $200,000–$350,000 in base salary, plus 0.1–0.7% equity.
• Experienced founders: Work alongside a team with a track record of building and selling companies.
• Cutting-edge technology: Work at the intersection of AI, video generation and creative applications.
• Visa sponsorship: The company can support eligible candidates who need sponsorship.

Location & Working Arrangements

This is an in-office position in San Francisco, California, five days per week. Candidates already based in the Bay Area and those willing to relocate are welcome to apply. Visa sponsorship is available.

Interested?

If you're an ML engineer who has trained modern generative-media models and successfully brought them into production, we'd love to hear from you.