Applied Research Scientist

Role Overview

We are seeking a Senior Research Scientist with deep expertise in Large Language Models (LLMs) and modern machine learning to help build and scale the Glance LLM, a domain-specific model focused on fashion intelligence, recommendation, and conversational commerce.

This role sits at the intersection of research, data, and systems, with responsibility for developing and evolving LLM capabilities from foundation model adaptation through to domain-specific pretraining and optimisation. You will contribute to the design of training pipelines, data strategies, and model architectures, with a focus on translating cutting-edge research into production-ready systems.

This is a hands-on role for someone with a strong research background who is motivated to see their work deployed at scale, shaping real-world user experiences.

Key Responsibilities
• Design and develop domain-specific LLM capabilities, including fine-tuning, continued pretraining, and alignment strategies
• Build and optimise end-to-end LLM pipelines, including data ingestion, tokenisation, training, evaluation, and deployment
• Lead research in representation learning for structured domains, particularly fashion, commerce, and multimodal data
• Develop approaches for retrieval-augmented generation (RAG), tool use, and grounded reasoning over product catalogues
• Work with large-scale datasets, including data curation, filtering, and synthetic data generation
• Contribute to model evaluation frameworks, including both automated benchmarks and human-aligned evaluation
• Collaborate with ML Engineers to productionise models, ensuring scalability, cost efficiency, and reliability
• Evaluate and integrate advances from leading models such as Gemma and Qwen
• Mentor junior researchers and contribute to a strong applied research culture

Required Qualifications
• PhD in Machine Learning, Natural Language Processing, Artificial Intelligence, or a related field
• Strong publication record in top-tier venues such as NeurIPS, ICML, ACL, EMNLP, or ICLR
• Deep understanding of transformer architectures, scaling laws, and LLM training methodologies
• Experience with fine-tuning, instruction tuning, or continued pretraining of large models
• Strong programming skills in Python, with experience in frameworks such as PyTorch
• Experience working with large-scale datasets and distributed training systems
• Ability to operate across research and engineering, with a focus on delivering practical impact

Preferred Qualifications
• Experience building or adapting domain-specific LLMs (e.g., for recommendation, commerce, or structured reasoning)
• Familiarity with retrieval-augmented systems (RAG) and vector databases
• Experience with model distillation, quantisation, or optimisation for deployment
• Exposure to multimodal models (text + image)
• Experience with evaluation design, including human alignment studies
• Contributions to open-source LLM frameworks or research artefacts

What You’ll Work On
• Building the Glance LLM, a domain-specific model optimised for fashion understanding and personalisation
• Developing systems that combine language understanding with product intelligence, enabling richer user interactions
• Designing scalable pipelines that balance model quality, cost, and latency
• Contributing to a long-term roadmap toward greater ownership of LLM capabilities, including pretraining and model optimisation
• Supporting integration with broader systems, including image generation and on-device AI

Why This Role Matters

At Glance, the LLM is not just a conversational interface, it is the intelligence layer that connects users, products, and experiences. The ability to deeply understand fashion, context, and user intent is central to delivering personalised, high-quality experiences at scale.

This role is critical in defining how that intelligence is built, adapted, and evolved over time, balancing rapid progress with long-term capability ownership and differentiation.

What We Offer
• Opportunity to work on cutting-edge LLM research with real-world deployment at scale
• A strong applied research environment balancing innovation and production impact
• Access to large-scale datasets and compute infrastructure
• The ability to shape a domain-specific AI platform from first principles
• Competitive compensation and growth opportunities