Artificial Intelligence and Machine Learning are moving rapidly from experimental projects into real engineering applications. However, successfully implementing AI involves considerably more than selecting an algorithm or training a model. Engineering teams must decide whether AI is appropriate for the problem, define measurable objectives, collect and manage suitable data, validate the resulting system and ensure that it can be deployed safely and responsibly.
To support engineers through these decisions, TechWorks has developed the Best Practices in AI Guide. TechWorks is now inviting engineers, AI practitioners, researchers, academics and other industry specialists to review the guide and provide practical feedback.
Who is the guide for?
The guide has been developed primarily for professional software engineers working within the electronic-systems industry. It assumes that the reader already has established engineering and software-development experience but may have limited previous knowledge of AI or Machine Learning. Rather than presenting AI as a completely separate engineering discipline, the guide is designed to help engineers transfer their existing skills into AI/ML development. Its objective is to provide a structured path from the initial consideration of an AI project through to the implementation and deployment of an AI-enabled system.
Five questions for an AI engineering project
The guide is organised around five fundamental questions that engineering teams should consider when planning and delivering an AI or Machine Learning project.
1. Should AI or Machine Learning be used?
Not every problem requires an AI solution. The first stage asks teams to establish the engineering case for using AI rather than a conventional algorithmic, rules-based or software approach. It also encourages early consideration of project feasibility, unacceptable AI risks and potential problems associated with the data that may be required.
2. How should the AI project be defined?
Once an engineering case has been established, the project needs clear goals, boundaries and measures of success. The guide considers how to define project objectives and key performance indicators while recognising that many AI systems operate statistically rather than deterministically. It also introduces the need for a wider project risk assessment before significant development work begins.
3. How should data be collected and managed?
The quality of an AI application is closely connected to the quality and suitability of the data used to develop it.
The guide therefore covers areas including:
- Data-collection planning
- Version control for code and data
- Documentation and logging
- Data exploration and cleaning
- Validation and testing
- Data storage and access
The objective is to help teams create a data process that is controlled, reproducible and appropriate for the intended application.
4. How should the AI/ML application be trained?
The training stage considers how engineering teams can choose an appropriate AI or Machine Learning approach and build a repeatable training pipeline. It addresses data preprocessing, testing, validation, potential biases, version control and the documentation of training decisions. The guide also encourages teams to revisit earlier assumptions when issues appear during training rather than treating the development process as a fixed sequence.
5. How should the AI application be deployed?
Deployment brings an AI model into a live or production environment. At this stage, teams must consider where the model will operate, how it will receive data, how updates will be controlled and how its performance will be monitored over time. The guide also discusses important deployment concerns such as unexpected inputs, model behaviour, security risks, data leakage and the continuing need for testing and human oversight.
Why is TechWorks seeking wider feedback?
A best-practice guide becomes more valuable when it reflects the experience of people working across different engineering disciplines, applications and industries.
TechWorks has already invested considerable work in developing the guide. It has now reached a stage where wider technical review and constructive critique can help identify:
- Missing engineering considerations
- Areas that require clearer explanation
- Assumptions that may not apply across every industry
- Additional examples or case studies
- Improvements to the task lists and project sequence
- Emerging safety, security or governance concerns
- Topics that may need to be expanded in future versions
Feedback is particularly welcome from people working in AI/ML development, embedded systems, electronic system design, functional safety, cybersecurity, data engineering, verification, regulation, research, and technical education.
What should reviewers consider?
Reviewers do not need to comment on every section of the guide.
Useful feedback could address questions such as:
- Is the technical guidance accurate?
- Is the proposed workflow practical for real engineering projects?
- Are any important project stages or decisions missing?
- Are the task lists clear and usable?
- Are safety, security, robustness and governance treated appropriately?
- Would additional examples make a section easier to apply?
- Is any part of the guide unclear, outdated or open to misinterpretation?
Comments based on practical project experience are especially valuable.
Review the guide and contribute
There are several ways to review the guide and provide feedback.
Comment directly in Google Docs
A simplified public Google Docs version has been created so that reviewers can read the guide in one place and leave comments against individual sections.
Open the comment-enabled Google Docs version
Read the published guide
The complete published version can be accessed through the TechWorks guide hub.
Read the published TechWorks Best Practices in AI Guide
Review or contribute through GitHub
Developers and technical contributors can review the source material through the public GitHub repository.
View the TechWorks Best Practices in AI GitHub repository
Contributors can also propose specific changes by submitting a pull request.
View or open a GitHub pull request
Send comments by email
Feedback can also be sent directly to:
William Jones, Head of AI, Embecosm
Email: william.jones@embecosm.com
Help strengthen practical AI engineering guidance
AI technologies and their applications continue to evolve, but strong engineering practice remains essential. Clear objectives, controlled data, reproducible development, rigorous validation, appropriate governance and continuous monitoring are all necessary if AI systems are to operate reliably in real applications.
By reviewing the TechWorks Best Practices in AI Guide, engineers and other specialists can help ensure that the guidance reflects practical industry experience and provides a useful foundation for teams beginning their AI/ML journey.
Written by : Mike Bartley
Mike started in software testing in 1988 after completing a PhD in Math, moving to semiconductor Design Verification (DV) in 1994, verifying designs (on Silicon and FPGA) going into commercial and safety-related sectors such as mobile phones, automotive, comms, cloud/data servers, and Artificial Intelligence. Mike built and managed state-of-the-art DV teams inside several companies, specialising in CPU verification.
Mike founded and grew a DV services company to 450+ engineers globally, successfully delivering services and solutions to over 50+ clients.
Mike started Alpinum in April 2016 to deliver a range of start-of-the art industry solutions:
Alpinum AI provides tools and automations using Artificial Intelligence to help companies reduce development costs (by up to 90%!) Alpinum Services provides RTL to GDS VLSI services from nearshore and offshore centres in Vietnam, India, Egypt, Eastern Europe, Mexico and Costa Rica. Alpinum Consulting also provides strategic board level consultancy services, helping companies to grow. Alpinum training department provides self-paced, fully online training in System Verilog, UVM Introduction and Advanced, Formal Verification, DV methodologies for SV, UVM, VHDL and OSVVM and CPU/RISC-V. Alpinum Events organises a number of free-to-attend industry events
You can contact Mike (mike@alpinumconsulting.com or +44 7796 307958) or book a meeting with Mike using Calendly (https://calendly.com/mike-alpinum-consulting).
Stay Informed and Stay Ahead
Latest Articles, Guides and News
Explore related insights from Alpinum that dive deeper into design verification challenges, practical solutions, and expert perspectives from across the global engineering landscape.
