Research Scientist – Video Generation

Pika

. Own RL-based post-training for video diffusion/flow-matching models .

Posted 9/24/2026full-timePalo Alto • California • United StatesJuniorMid-Level💰 $185,000 - $400,000 per yearWebsite

Core Competencies

Role fit

Core Competencies

Use this summary to align your resume positioning with the role.

Demonstrates expertise in reinforcement learning (RL) post-training for video models, with a strong focus on preference optimization, reward model development, and multi-node distributed training. Proficient in using PyTorch and familiar with video-specific challenges to enhance generative modeling outcomes.

Highest-signal resume keywords

Reinforcement Learning Post-TrainingPreference OptimizationPyTorch ProficiencyMulti-Node Distributed TrainingReward Model Development

ATS Keywords

Tailor your resume

Applicant Tracking System Keywords

Tip: use these terms in your resume and cover letter to boost ATS matches.

Hard Skills

Generative ModelingDiffusion ModelsFlow-Matching ModelsModel Improvement EvidenceDistillation ExpertiseAdversarial ApproachesEvaluation MetricsData Collection WorkflowsHuman Preference StudiesVideo-Specific Failure Modes

Soft Skills

CollaborationCommunication

Industry Keywords

Video GenerationTemporal DriftMotion RealismPhysics RealismVLM-as-Judge

Tech Stack

Tools & technologies

Node.jsPyTorch

About the role

Key responsibilities & impact

Own RL-based post-training for video diffusion/flow-matching models
Run RL post-training, including preference optimization and online RL against learned rewards, at multi-node scale
Build video reward models by defining evaluation dimensions, configuring preference data collection workflows, training and validating learned judges, and safeguarding against reward hacking
Own post-training evaluation using human preference studies and correlation with automated metrics
Distill RL-tuned models to efficient few-step samplers while preserving alignment gains
Collaborate with engineering and product teams
Shape real-time creative and agentic video platforms

Requirements

What you’ll need

2+ years hands-on research experience in post-training or generative modeling
RL or preference-optimization experience on generative models with evidence of model improvement
Strong grounding in diffusion or flow-matching models
Proficiency with PyTorch
Experience with multi-node distributed training
Experience developing reward models for visual generation, including VLM-as-judge or large-scale preference data collection (preferred)
Distillation expertise in distribution matching, consistency, or adversarial approaches, ideally for video models (preferred)
Familiarity with video-specific failure modes such as temporal drift, motion, and physics realism (preferred)
Ability to regularly work from the Palo Alto HQ
Authorization to work in the United States or ability to address US employment sponsorship requirements

Benefits

Comp & perks

Substantial equity in a high-growth startup
Full health benefits
401k matching
Significant growth opportunities
Flexible on-site/remote hybrid work arrangement