SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Twelve Labs builds multimodal AI models that understand video across sight, sound, and motion. The company has raised over $210M from top-tier investors including NEA, Amazon, NVIDIA, and Snowflake, and operates globally with offices in San Francisco, Seoul, New York, and London.
Pegasus is Twelve Labs' video-language model that transforms video into structured understanding by reasoning over visuals, speech, audio, and on-screen text. A core capability is Segment, a time-based metadata feature that allows customers to define segment types and metadata fields, with Pegasus returning relevant start/end times and structured metadata—enabling video to flow directly into search, archive, editing, compliance, and content-management workflows.
The Cognition Models team owns the models that turn video into structured understanding and reasoning: Pegasus and Jockey Core, the reasoning LLM behind Jockey (Twelve Labs' unified agentic system). The team focuses on multimodal systems with high instruction-following capability and complex hierarchical outputs, spanning training infrastructure from pre-training to RL, temporal segmentation, structured metadata extraction, large-scale inference and serving, data curation and evaluation pipelines, and building Jockey Core. The team ships products with real-world value using advanced compute including NVIDIA B300s.
In this role, you will drive technical direction for training infrastructure and training operations within Pegasus while remaining deeply hands-on in critical system design and implementation. You will own the design and evolution of scalable end-to-end training pipelines, with focus on reliability, reproducibility, efficiency, and fast iteration in large-scale distributed environments. You will lead technical decision-making across data curation workflows, training systems, evaluation pipelines, and ML infrastructure for multimodal model development. You will improve and automate the end-to-end training lifecycle so research ideas translate into robust systems integrated into production model development quickly and reliably. You will mentor engineers and raise the team's execution bar through strong technical judgment, design reviews, and hands-on collaboration. You will explore and adopt AI-assisted development tools such as Claude, Gemini, and GPT to improve productivity across coding, experimentation, debugging, and documentation.
