Real-Time Multimodal Vision Research Engineer

Acceler8 Talent; San Francisco (CA); On-site; USD 180,000 - 250,000; Full time

Job summary

Acceler8 Talent is partnering with a frontier AI research startup to recruit a Computer Vision Research Engineer. The role focuses on developing low-latency multimodal models for gaming environments with future robotics and human-AI collaboration applications.

You will work on research-grade video understanding models, build real-time perception systems, and optimize architectures for fast inference while exploring temporal representations across extended interactions.

Qualifications

• PhD in Computer Vision, Machine Learning or a related field.
• Strong publications at CVPR, ICCV, ECCV, NeurIPS, ICLR or ICML.
• Deep experience with VLMs, video understanding or multimodal transformers.
• Strong Python/PyTorch research engineering experience.
• Familiarity with recent VLM/VLA research and real-time visual perception.

Responsibilities

• Research and develop video understanding models, VLMs and multimodal architectures.
• Build real-time perception systems for recognising actions, objects and spatial relationships.
• Develop temporal representations for understanding events across extended interactions.
• Train and evaluate models using demonstrations, synthetic data and interactive environments.
• Optimise visual models for low-latency inference and real-time interaction.

Skills

Python/PyTorch

VLMs

Multimodal transformers

Education

PhD in Computer Vision or ML

Job description

Acceler8 Talent is partnering with a frontier AI research startup to recruit a Computer Vision Research Engineer. The role focuses on developing low-latency multimodal models for gaming environments with future robotics and human-AI collaboration applications.

You will work on research-grade video understanding models, build real-time perception systems, and optimize architectures for fast inference while exploring temporal representations across extended interactions.