Data Scientist

Mott MacDonald is a global engineering, management, and development consultancy committed to delivering impactful work that shapes the future.

We are a team of over 20,000 experts working across the world in more than 50 countries. We are proud to be part of an ever-changing global industry, delivering transformative work that’s defining our future. It’s our people who power that performance. As an employee-owned business, we invest in creating a space for everyone to feel safe and valued and empowered with the right tools and support. 

Whether you want to pursue excellence in your specialism or broaden your experience with flexible roles across our business, you’ll be connected to a community of global experts championing you to be your best. Join us and shape your story with Mott MacDonald, where everyone has the opportunity to be brilliant.

We help our client’s put information at the heart of what they do and derive ever greater value from it. Our focus is the provision of outcome focused, game-changing results. This is powered by a unique combination of domain plus digital expertise, harnessing 120 years of engineering legacy alongside the power of cutting-edge technology, from both Mott MacDonald and our partners. By ‘connecting innovation to outcomes' we can continue to enable better outcomes across the planning, delivery, and operation of our client's asset base. 
 

We are looking for a Data Scientist with deep expertise in Computer Vision to lead the development of intelligent systems that can interpret the physical world. Whether it’s using satellite imagery to track carbon sequestration, deploying Vision Transformers for structural health monitoring, or leveraging VLLMs to query vast libraries of site photos, your work will turn pixels into actionable engineering insights.

Your Impactful Role 

• Visionary Architect: Own the end-to-end lifecycle of Computer Vision projects, from data acquisition and model selection to deployment within our digital twin platform. 

• Next-Gen Modelling: Move beyond traditional CNNs to implement state-of-the-art Vision Transformers (ViTs) and Multimodal VLLMs for complex scene understanding. 

• Intelligent Inspection: Design systems that automate the detection and classification of infrastructure defects, site safety hazards, and environmental changes. 

• Multimodal Reasoning: Build and deploy Vision-Language Models (e.g., LLaVA, CLIP) that allow engineers to "chat" with their visual data using natural language. 

• Scalable Inference: Implement high-performance serving architectures using vLLM and Azure AI to process high-resolution imagery and video streams at scale. 

Technical Excellence You Bring 

1. Advanced Computer Vision & Multimodal AI 

• Vision Core: 5+ years of experience in computer and deep learning with strong foundation of mathematical concepts. 

• Transformer-Based Vision: Deep hands-on experience with Vision Transformers (ViT, Swin, DeiT) and Detection/Segmentation/Image generation Transformers (e.g., DETR, Mask2Former, SAM).  

• Vision-Language Mastery: Proven experience working with VLLMs (Vision Large Language Models) like LLaVA, Phi-Vision, or BLIP for visual question answering and image captioning. 

• Traditional & Deep CV: Strong foundations in Object Detection, Instance Segmentation, and Image Registration using libraries like Detectron2 or YOLOv10+. 

• Zero-Shot Learning: Expertise in using CLIP-based contrastive learning for flexible, open-vocabulary visual recognition. 

2. Machine Learning & Engineering 

• ML Core: 5+ years of experience in the ML domain with a strong grasp of mathematical modelling and statistical analysis. 

• Deep Learning Frameworks: Mastery of PyTorch (preferred) or TensorFlow for training and fine-tuning large-scale vision models. 

• Software Craftsmanship: Proficiency in Python with a focus on clean code, design patterns, and scalable architecture. Prior Exposure to end-to-end software development in enterprise environment.

3. Deployment & Cloud (Azure)

• Inference Optimization: Should have experience using vLLM for serving multimodal models. Nice to have knowledge of quantization (e.g., AWQ, GPTQ) for efficient vision-at-the-edge. 

• Cloud Infrastructure: Extensive experience deploying vision pipelines on Azure ML, including managed endpoints and large-scale data processing. 

• Ops & Data: Knowledge of MLOps for vision—handling high-resolution (4K) datasets, data labelling strategies, and model monitoring in production. Nice to have knowledge on cloud data points, 360 render graphics. 

Desired Attributes 

• Domain Passion: An interest in applying AI to engineering, geospatial data science, or urban planning. 

• Strategic Communication: The ability to explain complex multimodal AI concepts to project managers and structural engineers. 

• Enthusiastic Innovator: A "hacker" mindset—staying current with the latest CVPR/NeurIPS research and bringing state-of-the-art papers into production. 

Why Mott MacDonald? 

We are a 16,000-strong network of experts, but we operate with the agility of a tech-first consultancy. 

• Purpose-Driven Work: Every model you build helps solve a global challenge, from clean water access to smarter transportation.

• Professional Growth: We invest in our people with continuous learning paths in AI, Data Science, and Leadership. 

Inclusion: We are an equal-opportunity employer that values the diversity of thought you bring to our "Connected Thinking." 

Flexible working 

At Mott MacDonald, we support our staff to create work-life balance which works for them and welcome candidates looking for career flexibility. We are open to discussing flexible working at interview stage.

Our benefits package is designed to enhance your experience:

• Agile working
• Critical illness and compassionate leave
• Paternity Leave
• Group term life insurance, and Group medical insurance coverage
• Career mobility options
• Short and Long-term global employment opportunities
• Global collaboration and knowledge sharing