Machine Learning R&D Engineer (KTP Associate) (London)

Join to apply for theMachine Learning R&D Engineer (KTP Associate)role atBuro Happold

Join to apply for theMachine Learning R&D Engineer (KTP Associate)role atBuro Happold

In partnership with Buro Happold, Birmingham City University are looking to appoint a high calibre graduate (graduated within the last five years) as a Machine Learning R&D Engineer (KTP Associate).

As a Machine Learning R&D Engineer Associate, you will develop Machine Learning (ML) solutions that will innovate and transform key engineering activities in Buro Happold. You will work closely with Buro Happold’s computational team to build reusable AI/ML datasets, develop and optimise ML models, facilitate their deployment within the company, and co-define data-driven solutions that have the potential to revolutionise the Architecture, Engineering and Construction (AEC) industry.

This role presents an exciting opportunity to work in collaboration with leading academics at Birmingham City University, to apply knowledge and technical innovation, delivered on site at the company.

The Machine Learning R&D Engineer (KTP Associate) should have a minimum 2.1 University qualification in a relevant subject area and graduated within the last five years.

This Knowledge Transfer Project (KTP) is co-funded by a grant from Innovate UK and Buro Happold Limited. It is therefore essential you understand the fundamentals of the KTP collaboration between a UK business and a University works to deliver benefits for each (the company, the university, and the graduate) For more information please go to www.ktp.org.uk

Personal Training & Development Budget:

The Machine Learning R&D Engineer (KTP Associate) will have access to a wider range of benefits including a personal development budget of£3,000to upskill during the project.

The successful candidate will be employed by Birmingham City University and seconded to work full-time onsite at Buro Happold Limited to deliver the 18-month KTP project in partnership Birmingham City University and Buro Happold Limited.

Buro Happold Limited:

Buro Happold Limited is an international, integrated consultancy of engineers, designers and advisers (about 3000 employees worldwide) that designs and delivers a wide range of construction projects (recent award-winning projects include Battersea Power Station in London and the K64 Keflavik Airport Area Masterplan). Buro Happold covers several domains of expertise, including structural engineering, façade design, MEP engineering, and many others. Buro Happold has a long history of developing and applying innovative computational solutions, like the open-source frameworkBuildings and Habitats object Model (BHoM)and is now investing into AI/ML innovation to increase their design efficiency and transform key workflows, ultimately sustaining BH’s position as one of the key leaders in the AEC sector. For more information, please go to www.burohappold.com.

The successful candidate will have full access to Birmingham City University’s resources such as offices, labs, and library to complete the KTP project (a project workplan written with KPIs and outcome deliverables has been written).

The Machine Learning and R&D Engineer (KTP Associate) will be supervised and mentored by both a lead academic and academic supervisor academics from Birmingham City University’s College of Computing and Digital Technology within the Faculty of Computing, Engineering and Built Environment (CEBE) as well as a company supervisor located at Buro Happold Limited whose aim is to assist the Machine Learning and R&D Engineer (KTP Associate) to deliver the knowledge into Buro Happold Limited and successfully deliver the 18-month KTP project on behalf of Buro Happold and Birmingham City University.

Main Duties and Responsibilities:

The Machine Learning and R&D Engineer (KTP Associate) duties and responsibilities will include the following:

• Understand the main workflows of BH. This includes interfacing with domain experts to understand some basic concepts of the different domains of work (e.g. Structural engineering, façade design, electrical engineering) which will be necessary to perform the ML work.
• At this stage, some main use cases for ML R&D have already been defined. The candidate will need to understand the collected use cases and how they apply to BH workflows. This includes talking to domain experts and making sure that all relevant use case information is available, identifying possible gaps, providing recommendations where needed.
• Define a ML R&D strategy for the collected use cases. This includes performing literature review, understanding challenges, and proposing realistically achievable goals with a clearly defined plan and timeline.
• Help develop reusable datasets suitable for training ML models, using Buro Happold’s extensive data sources. This includes co-defining labelling processes according to the developed KG and coordinate with domain experts that will perform the labelling.
• Select and fine-tune computer vision models leveraging the developed datasets to identify and classify elements. This includes being able to leverage existing models/architectures, or develop new architectures, depending on the use case and the resources available at specific times (from lightweight RCNNs to more complex models like YOLO).
• Select appropriate modelling techniques (e.g. few-shot learning) and data augmentation strategies when appropriate to adapt the developed solution to low-data contexts.
• Contribute to the development of an Ontology/Knowledge Graph (KG) to represent key engineering concepts.
• Train/develop graph ML models with an approach similar to scene graphs leveraging the KG for the collected use cases.
• Be proactive in highlighting blockers and requesting support or extra data when needed, as appropriate.
• Collaborate with other Computational Team’s engineers to facilitate the deployment of the models within BH’s workflows. This may include minor front-end and MLOps tasks.
• Write periodic reports to show progress and participate in team planning/review activities with both BH and BCU.
• Co-author papers and participate in conferences and dissemination activities within and outside BH and BCU.

Competencies, Skills and Experience:

The Machine Learning and R&D Engineer (KTP Associate) will hold a relevant degree. The candidate should have a good first degree and post-graduate degree in Computer Science or Data Science, or a related discipline, excellent programming skills, good communication skills and experience of successfully working as part of a team.

We expect the post holder to have the following requirements:

• Minimum of 2.1 and ideally a Master’s degree in Computer Science, Data Science, or a related field (e.g., artificial intelligence), or equivalent industry experience.
• Ability to perform literature review, understanding scientific papers, drawing insights and ideas for implementation. Having previous publications is a plus.
• Experience in designing, developing, and implementing computer vision models and algorithms.
• Proficiency in Python and its standard coding practices and common libraries.
• Experience with ML models and knowledge of at least one ML framework (PyTorch preferred).
• Experience in data pre-processing techniques, feature engineering, and model evaluation metrics.
• Proven ability to manipulate, query and visualise data and training/evaluation results (using e.g., Pandas, Matplotlib, Seaborn).
• Proven understanding of Information Extraction and Retrieval techniques.
• Proven understanding of NLP and large language models.
• Proven understanding of database concepts (differences between graph/relational/non-relational).
• Proven familiarity with cloud computing platforms (e.g., AWS, Azure, GCP), Azure preferred.
• Experience with version control tools (i.e. Git). Experience with dataset and model versioning tools is a plus.
• Experience with Linux.

Highly Desirable:

• Experience with Graph Deep Learning, scene graphs, or similar techniques.
• Experience with Ontologies/KGs.
• Experience with Graph Databases (any, triple store or property graph).
• Previous R&D and/or research experience.
• Basic knowledge of Windows, and willing to learn basic knowledge of some Microsoft-based tools needed for data acquisition and deployment (mainly SharePoint).
• Experience with Azure or other cloud services.
• Experience with dataset versioning tools (e.g. DVC)
• Experience with ML models versioning tools (e.g. MLFlow)
• Understanding of data structures, algorithms, OOP concepts.
• Understanding of basic software engineering design patterns.
• Interest in AEC or previous experience or research with the AEC industry.

Personal Skills:

• Proactive in identifying risks, blockers and communicating requests for support.
• Ability to convey technical concepts to both technical and non-technical stakeholders.
• Clear communication and presentation skills, both written and verbal.
• Ability to work both independently and within a team.
• Willingness to understand domain-specific use cases and the needs of key non-technical stakeholders.
• Problem-solving and critical-thinking abilities. Does not overfocus on a specific problem, keeps the big picture in mind.
• Able to alternate between different tasks. Knows when to stop and change activity.
• Willing to learn and to in