Tech Lead Manager, Jockey Core

About the role

You will build and lead the founding team responsible for Jockey Core, the reasoning LLM at the center of Jockey that decomposes queries, decides what to retrieve and segment, and reasons over results into actionable answers. You will own end-to-end technical direction for the model stack, balancing research innovation with production constraints such as latency, cost, and reliability. You will stay deeply hands-on as a practitioner while managing people, processes, and cross-functional alignment to deliver a scalable agentic product. You will define and drive the model efficiency roadmap, including pruning, quantization, and distillation, and partner closely with infrastructure and product teams to ensure smooth production serving. You will lead data curation, evaluation pipelines, and continuous improvement cycles so that Jockey Core quality compounds with every release. You will act as the single point of ownership for critical decisions that span algorithms, serving systems, and product requirements. You will communicate progress and trade-offs clearly to both technical and non-technical stakeholders, ensuring alignment with TwelveLabs' global product goals.

Key facts

What you'll do

Lead the founding engineering and research team for Jockey Core, defining hiring plans, career growth, and technical execution.

Own the end-to-end Jockey Core roadmap, from model and engine selection through model-efficiency optimizations to production serving and scalability.

Partner with the Perception Models team to align retrieval, segmentation, and embedding strategies with Jockey Core requirements.

Drive model-efficiency initiatives such as pruning, quantization, and distillation to meet latency and cost targets without sacrificing quality.

Define and iterate on evaluation frameworks and data curation pipelines to ensure robust, measurable progress in reasoning and agentic capabilities.

Lead the integration of model improvements into production serving systems, coordinating with infrastructure and platform teams.

Establish clear product requirements and success metrics for Jockey Core, translating them into technical milestones and delivery plans.

Conduct hands-on technical work, including experiments, debugging, and code reviews, to unblock the team and set technical standards.

Build and maintain strong working relationships with product managers, researchers, and leaders across TwelveLabs to prioritize work effectively.

Communicate roadmap status, risks, and trade-offs to both technical and non-technical audiences in a fast-paced, global environment.

Ensure delivery against milestones by coordinating cross-functional dependencies and managing timelines and resources.

Promote a culture of openness, rigor, and continuous learning within the team to attract and retain top talent.

Represent Jockey Core in internal and external discussions, ensuring consistent messaging and alignment with company strategy.

Guide technical decision-making for the model stack, balancing innovation with practical constraints of production environments.

Requirements

Must have a strong background in machine learning or related fields with hands-on experience in model training, optimization, or inference.

Have demonstrated experience in managing engineering teams and delivering complex technical products at scale.

Bring deep expertise in large language models, including architecture decisions, training techniques, and inference optimization.

Show a proven track record of balancing research advances with production constraints such as latency, reliability, and cost.

Possess excellent communication skills to align cross-functional stakeholders and translate strategic goals into technical execution.

Have experience working with multimodal data and agentic workflows relevant to video and image understanding.

Be comfortable working in a fast-paced, rapidly evolving startup environment with ambiguous problems and shifting priorities.

Demonstrate strong ownership and leadership qualities, with the ability to make high-impact decisions under uncertainty.

Nice to have

Experience with building and serving large-scale AI inference systems in production.

Background in multimodal models, video understanding, or related domains.

Experience with model compression techniques such as pruning, quantization, and distillation.

Familiarity with agentic AI frameworks and evaluation methodologies for reasoning tasks.

Practical notes

This is a full-time position based in Seoul, South Korea.

Travel requirements, visa sponsorship details, and application deadlines are not specified in this listing.