Member of Engineering (Multimodality – Research Lead)

About the role

This position is based in South Africa and leads the development of multimodal capabilities for next‑generation AI systems. You will shape a new research area that enables advanced models to understand and interact with visual information, working with large‑scale training infrastructure and a collaborative research team.

Key responsibilities

• Own the multimodality research roadmap and drive adoption across AI models.
• Build and launch initial image‑input capabilities for advanced coding and agentic systems.
• Define the evolution from adapter‑based approaches to native multimodal architectures.
• Design, implement, and scale experiments across the full research lifecycle.
• Collaborate with cross‑functional teams on evaluation, architecture, data, and post‑training.
• Create custom datasets and evaluation strategies for multimodal software‑engineering capabilities.
• Contribute to establishing a new research function from the ground up.

Required profile

• Strong machine‑learning expertise with a research‑driven mindset.
• Proven experience building large‑scale AI systems and production‑grade vision‑language models.
• Ability to work independently in ambiguous, fast‑paced environments.
• Demonstrated leadership or 0‑to‑1 building experience with broad ownership.
• Excellent experimental design, execution, and analysis skills.

Required skills

• Python programming
• Vision‑Language Models (VLMs)
• Transformers and large language model fundamentals
• Distributed training at scale
• Machine learning system development