Machine learning engineer

Knowledge

K1: The purpose, methodologies and applications for ML AI solutions such as Machine Learning, Computer (Machine) Vision, batched learning systems, Robotics, Generative Transformer Models and Natural & Large Language Processing (NLP and LLMs) Models. Back to Duty

K2: The stages of the machine learning lifecycle. Including establishing the model objectives, data preparation, building and training the model, ML problem framing, testing and evaluating the model using the preferred framework, deploying the modelling and monitoring, maintaining and updating the model using process frameworks such as Quality Assurance and either online, continuous (CLS) or batched learning systems. Back to Duty

K3: Vulnerabilities related to confidentiality, authentication, non-repudiation, service integrity, network security, planned or unplanned adversarial danger, threat or attack, host OS security, physical security and the implications and preventative mitigations for these at all stages of the machine learning lifecycle. Back to Duty

K4: Project Management methodologies and techniques for machine learning activities such as CRISP-ML Cross Industry Standard Process. Back to Duty

K5: Differences and applications of machine learning methods, and models such as: supervised learning; semi supervised learning; unsupervised learning; natural language processing ; reinforcement learning; ensemble learning; predictive using tools for experiment tracking, orchestration, versioning, deployment and monitoring. Back to Duty

K6: The risks that might occur for example bias, security, quality or over fitting in the product lifecycle during building, testing and through to deployment of ML models in the live environment. Back to Duty

K7: How to identify and select the performance metrics of the proposed model in the context of the business need. Back to Duty

K8: The processes used to identify variables and features that can impact stability of model performance during testing and when applying changes to existing models in the live environment. Back to Duty

K9: The importance of feature engineering, selection and pre-processing in effective machine learning. Back to Duty

K10: Machine learning implementation principles for data engineering solutions including quality, security, efficiency, validity, training, testing and tuning. Back to Duty

K11: How machine learning methods are applied to maximise the impact to the organisation. Back to Duty

K12: Deployment approaches for new data pipelines and automated processes. Back to Duty

K13: Data and information security standards, ethical practices, policies and procedures relevant to data management activities such as data lineage, data retention and metadata management. Back to Duty

K14: Change management processes for ML solutions; recording and logging change using appropriate tools and documentation. Back to Duty

K15: The implications of data types ( for example variety, quality, formats) on security, scalability, governance for ML and or AI infrastructure, and cost of local, remote or distributed solutions such as cloud and other SaaS and PasS ML/AI providers. Back to Duty

K16: How to use programming languages, integrated development environments and modern machine learning libraries. Back to Duty

K17: Principles for engineering environmental sustainable ML solutions, that support organisational strategies and objectives for environmental sustainability. Back to Duty

K18: The relationship between mathematical principles and core techniques in machine learning and data science within the organisational context. Back to Duty

K19: How to solve problems and evaluate software solutions via analysis of test data including synthetic data and results from research, feasibility, acceptance and usability testing. Back to Duty

K20: Sources of error and algorithmic bias, including how they may be affected by choice of dataset and methodologies applied using practices such as Explicability and Explainable AI (XAI). Back to Duty

K21: The methods and techniques used to communicate concepts and messages to meet the needs of the audience, adapting communication techniques accordingly. Back to Duty

K22: Approaches and strategies to stakeholder engagement including engagement with the end user Back to Duty

K23: How machine learning and data science techniques support and enhance the work of other members of the team. Back to Duty

K24: Concepts of data governance, including regulatory requirements, data privacy, security, trustworthiness and quality control. Back to Duty

K25: Legislation, regulation, governance and guidance assurance frameworks for example AREA or SAFE D and their application to the safe interoperable use of data, machine learning and artificial intelligence. Back to Duty

K26: The ethical aspects associated with the use and collation of data and machine learning models. Back to Duty

K27: What the cyber security culture in an organisation is, and how it may contribute to security risk. Back to Duty

K28: How to identify trends and emerging technologies to ensure knowledge is up to date with new developments in machine learning and AI such as AI embedded within tooling. Back to Duty

K29: How own role supports ML solutions in accordance with organisational strategies, business requirements, Corporate Governance Principles, Social Corporate Responsibilities, legal regulations and Ethical Practices. Back to Duty

K30: AI based approaches, including those provided by third-party vendors’ (Application Programming Interfaces), into existing and new processes. Back to Duty

K31: Software development best practices; for example, software testing, version control, continuous integration and continuous delivery. Back to Duty

Skills

S1: Assess vulnerabilities of the proposed design, to ensure that security considerations are built in from inception and throughout the development process. Back to Duty

S2: Translate business needs and technical problems to scope machine learning engineering solutions. Back to Duty

S3: Select and engineer data sets, algorithms and modelling techniques required to develop the machine learning solution. Back to Duty

S4: Apply methodologies and project management techniques for the machine learning activities. Back to Duty

S5: Create and deploy models to produce machine learning solutions. Back to Duty

S6: Document the creation, operation and lifecycle management of assets during the model lifecycle. Back to Duty

S7: Apply techniques for output model testing and tuning to assess accuracy, fit, validity and robustness. Back to Duty

S8: Assess system vulnerabilities and mitigate the threats or risks to assets, data and cyber security. Back to Duty

S9: Refine or re-engineer the model to improve solution performance. Back to Duty

S10: Apply techniques for monitoring models in the live environment to check they remain fit for purpose and stable. Back to Duty

S11: Consider the associated regulatory, legal, ethical and governance issues when evaluating choices at each stage of the data process. Back to Duty

S12: Apply machine learning and data science techniques to solve complex business problems. Back to Duty

S13: Track and test continual learning models. Back to Duty

S14: Analyse test data, interpret results and evaluate the suitability of proposed solutions both new and inherited models, considering current and future business requirements. Back to Duty

S15: Identify, consider and advocate for ML solutions to deliver an environmental and operational sustainable outcome. Back to Duty

S16: Transition prototypes into the live environment. Back to Duty

S17: Complete audit activities in compliance with policies, governance, industry regulation and standards. Back to Duty

S18: Consider the risks with using digital and physical supply chains. Back to Duty

S19: Ensure the model capacity is scaled in proportion to the operating requirements. Back to Duty

S20: Support the evaluation and validation of machine learning models and statistical evidence to minimise algorithmic bias being introduced. Back to Duty

S21: Monitor data curation and data quality controls including for synthetic data. Back to Duty

S22: Identify and select the machine learning or artificial intelligence platform architecture and specific hardware, to contribute to solving a computational problem using allocated resources. Back to Duty

S23: Identify and embed changes in work to deliver sustainable outcomes. Back to Duty

S24: Monitor model data drift, using performance metrics to ensure systems are robust when moving outside of their domain of applicability. Back to Duty

S25: Develop a process to decommission assets in line with policy and procedures. Manage current and legacy models in line with industry approaches. Back to Duty

S26: Undertake independent, impartial decision-making respecting the opinions and views of others in complex, unpredictable and changing circumstances. Back to Duty

S27: Coordinate, negotiate with and manage expectations of diverse stakeholders suppliers and multi-disciplinary teams with conflicting priorities, interests and timescales. Back to Duty

S28: Produce and maintain technical documentation explaining the data product, that meets organisational, technical and non-technical user requirements, retaining critical information. Back to Duty

S29: Create and disseminate reports, presentations and other documentation that details the model development to confirm stakeholder approval for handover to implementation. Back to Duty

S30: Comply with equality, diversity, and inclusion policies and procedures in the workplace. Back to Duty

S31: Horizon scan to identify new technological developments that offer increased performance of data products. Back to Duty

S32: Apply Machine Learning principles and standards such as, organisational policies, procedures or professional body requirements. Back to Duty

S33: Integrate AI-based approaches, including those provided by third-party vendors’ Application Programming Interfaces, into existing and new processes. Back to Duty

S34: Proactive identification of the potential for automation for example through AI solutions embedded within tooling. Back to Duty

Behaviours

B1: Uses initiative and innovation concerning new and emerging technologies through self directed learning and horizon scanning. Back to Duty

B2: Takes personal responsibility and prioritises sustainable outcomes in how they carry out the duties of their role. Back to Duty

B3: Acts inclusively when collaborating with people from technical and non-technical backgrounds. Contributing to knowledge sharing, management and empowerment across the broader team. Back to Duty

B4: Acts with integrity, giving due regard to legal, ethical and regulatory requirements. Back to Duty

B5: Operates in settings of technical complexity and uncertainty. Back to Duty