Staff ML Research Engineer, Multimodal Structure & Marengo

The role

You will set the technical direction for next-generation multimodal models and own the end-to-end development process. This includes research strategy, data architecture, training systems, production model APIs, and evaluation frameworks.

What you will do

• Set the technical direction for multimodal structure
• Define architecture, training, and data strategy for multimodal embedding models
• Own end-to-end model development from research to production
• Architect and optimize large-scale training infrastructure
• Own production model APIs from packaging to inference optimization
• Drive data strategy through large-scale curation and quality systems
• Define evaluation methods and quality standards
• Define the interface between structure and semantics
• Drive cross-functional alignment with Agent, Search, Product, and Infrastructure teams
• Raise the research engineering bar through design review and technical mentorship

Required qualifications

• 7+ years of industry experience in computer vision, video understanding, or multimodal learning
• Ability to drive ambiguous research problems to concrete solutions
• Deep expertise in large-scale distributed model training (kernel optimization, FSDP, or similar)
• Experience building and operating production model APIs for large-scale ML systems
• Deep expertise in video understanding, multimodal representation learning, or foundation model development
• Experience building end-to-end systems connecting multimodal structure with embeddings and retrieval
• Proven end-to-end ownership of building, deploying, and iterating in production
• Strong proficiency in Python and PyTorch
• Evidence of both research depth and engineering impact
• No disqualification for overseas travel

Nice to have

• Experience training models at billion-parameter scale
• Experience with training operations: pipeline reliability, monitoring, fault tolerance, and cost optimization
• Experience with large-scale data curation and data quality systems
• Experience modeling temporal, spatial, or hierarchical structure across video and other multimodal content
• Experience with temporal video understanding or multimodal video modeling
• Deep experience optimizing training and inference systems for throughput, latency, GPU efficiency, and scale
• Track record of technical leadership

Benefits

• Hybrid work with autonomy and collaboration
• Latest MacBook and 700,000 KRW worth of remote work equipment
• Equipment replacement every 3 years
• Unlimited LLM token support
• 1.4 million KRW annual self-development fund for lectures, conferences, and memberships
• English education and global buddy programs
• Taxi fare support for night and weekend commutes
• Corporate card worth 7.2 million KRW annually for meals and transportation
• Office snack bar with snacks, coffee, and seasonal fruit
• Dinner provided after 7 PM when working in the office
• Annual health checkup for employee and one family member
• Group insurance (accident, dental, or family accident)

Before you apply

• No disqualification for overseas travel
• 3-month probation period applies

About the company

TwelveLabs builds the intelligence layer for video data with multimodal AI models. The company is headquartered in San Francisco with offices in Seoul, New York, and London.