Research Scientist (post-training)

GenmoSan Francisco HQResearch

Apply on company site

Opens the official Genmo careers page

Salary

Not listed

Job type

Full-time

Posted

28 Sept 2026

Status

Open · checked 7d ago

About this role

This Research Scientist (post-training) position at Home in San Francisco HQ involves leading alignment and post-training initiatives for large-scale video generation models. This full-time role suits an experienced machine learning researcher with a strong background in reinforcement learning and generative artificial intelligence.

Role Overview

Home operates as a research lab building advanced video world models designed to simulate and understand the physical world to unlock the right brain of AGI. The organization seeks an exceptional Research Scientist to center on post-training and alignment techniques for diffusion-based video generation architectures. In this capacity, the professional ensures that large-scale models generate reliable, safe, and physically accurate outputs that align with human values and preferences.

Key Responsibilities

Direct research initiatives centered on alignment and post-training methodologies for video generation models to boost reliability, quality, and human intent adherence.

Build and execute reinforcement learning from human feedback and supervised fine-tuning pipelines specifically for video generation.

Formulate strong evaluation frameworks to gauge model output quality, safety, and overall alignment.

Build and refine data collection pipelines dedicated to human preferences and feedback.

Plan and execute experiments to test alignment methodologies and understand their scaling behaviors.

Partner with multidisciplinary teams to embed alignment upgrades straight into production systems.

Monitor the latest academic literature across generative AI and model alignment.

Guide junior researchers while championing responsible AI development practices.

Align closely with product units to ensure model capabilities improve alongside alignment upgrades.

Requirements

Earned Ph.D. within Artificial Intelligence, Computer Science, Machine Learning, or a closely related technical discipline.

Documented publication history in premier conferences like ICLR, ICML, or NeurIPS focusing on generative models, alignment, or reinforcement learning.

Deep hands-on background in optimizing and deploying large-scale training pipelines via PyTorch.

Comprehensive grasp of reinforcement learning methodologies, particularly RLHF.

Practical history running distributed training infrastructure and large-scale experiments.

Demonstrated success in building stable evaluation frameworks.

Exceptional communication capabilities to convey intricate technical information to varied groups.

Robust software engineering skills combined with experience navigating complex shared codebases.

Nice to Have

Background working with diffusion models or alternative generative architectures.

Prior experience fine-tuning large language models or generative systems.

Exposure to human feedback data collection and annotation frameworks.

Strong aesthetic judgment alongside knowledge of video quality evaluation.

Acquaintance with alternate alignment methods such as debate or constitutional AI.

History of productive teamwork alongside product divisions.

Familiarity with human evaluation design and perceptual quality metrics.

Active contributions to open-source initiatives within generative AI or alignment.

Skills and Tools

PyTorch

Reinforcement Learning / RLHF

Distributed Training Systems

Diffusion Models

Supervised Fine-Tuning

About the Company

Home, known as Genmo, builds sophisticated video world models focused on advancing physical intelligence and allowing robots to interact with a changing world. The business evaluates candidates without regard to religion, race, age, color, sex, national origin, disability, sexual orientation, veteran status, or other legally protected characteristics. The organization functions as an E-Verify participant.

How to Apply

Interested candidates should visit the official company careers page at genmo.ai to submit their application for this San Francisco HQ position.

Skills and tech stack

PyTorch Reinforcement Learning Machine Learning Diffusion Fine-tuning Generative AI

Questions candidates ask

Where is this job located?

This full-time position is based at the company headquarters in San Francisco, California.

What core technical skill is required?

Candidates must have extensive experience implementing and optimizing large-scale training pipelines using PyTorch.

What educational background is required?

Applicants must hold a Ph.D. in Computer Science, Artificial Intelligence, Machine Learning, or a closely related field.

What primary research focus does this role have?

The position focuses on alignment and post-training techniques for large-scale video generation models.

Is remote work permitted for this role?

The role is based in the Bay Area, and candidates are expected to be located near the area or be open to relocation.

Ready to apply?

The application is on the official Genmo careers page.

Apply on company site