How Enterprises Should Approach Training for Engineers in the AI Era

AI is no longer a pilot program sitting outside core engineering workflows. It’s embedded in how software gets built, tested, and shipped, and organizations have already invested heavily in making that shift. Gartner forecasts global AI spending will reach nearly $1.5 trillion in 2025 and top $2 trillion in 2026, with capital flowing into infrastructure, tooling, and integration at a pace that has outrun most organizations’ ability to build the skills required to use it well. The expectation now is that AI must deliver measurable business outcomes. Yet most enterprise upskilling efforts remain fragmented into a course here or a certification there, with no coherent strategy guiding them. That strategic gap is expensive.

Training for engineers in the AI era must be an enterprise capability built on a deliberate roadmap. This article covers the four essentials that will get you there: conducting a skills audit, prioritizing the right competencies, choosing the right formats, and proving ROI to leadership.

Overview

• AI is now embedded in daily engineering workflows across most organizations. Training that covers only tool proficiency leaves trust and governance gaps that create real operational risk.
• A structured skills audit using recognized frameworks helps organizations target investment where it matters most.
• Competency priorities should be layered: AI literacy for all engineers, with deeper specialization mapped to specific roles.
• ROI measurement must go beyond completion rates to track behavioral change and business outcomes.

The urgency of rapid AI adoption in today’s era

90% of software development professionals now report using AI in their workflows, which makes closing skills gaps across engineering teams more urgent than ever. Enterprises that rely on informal, individual-driven upskilling will struggle to keep pace with teams that invest in structured, strategically aligned learning programs. Organizations that treat training as a future problem are already falling behind the teams that started twelve months ago.

Why does AI demand a new approach to training?

AI, automation, agents, and data-driven workflows are reshaping engineering roles across every discipline, but the productivity gains are balanced by very real business risks. For example, delivery cycles slow when engineers can’t effectively review AI-generated code, and products may fail compliance obligations if teams aren’t trained on emerging regulatory requirements.

More importantly, AI amplifies existing engineering weaknesses just as much as it amplifies strengths. DORA’s Nathen Harvey has described this dynamic in conversation with Sam Newman: AI acts as an amplifier, so teams with strong practices get faster and teams with weak ones compound their problems at speed. Organizations with poor code review practices ship AI-generated vulnerabilities faster than they did before.

In this context, one-time training events are insufficient because the underlying challenges are always evolving. Continuous learning, closely tied to how teams actually build and ship software, is the only approach that holds.

How to identify skill gaps across engineering teams

A structured skills audit helps leadership move from assumptions to evidence when prioritizing engineer training programs.

Specifically, focus on the following:

• Create an inventory of roles and responsibilities across engineering teams to establish a clear baseline.
• Map day-to-day tasks to each role to capture how work is actually executed rather than relying on job descriptions.
• Benchmark tasks against AI-era skill requirements, ensuring teams can operate with AI in the loop.
• Use frameworks like NIST’s NICE Framework to standardize skills, tasks, and knowledge areas, especially for security-adjacent roles.
• Run anonymous team surveys to capture self-reported gaps and confidence levels across key competencies.
• Analyze codebases and workflows to identify practical gaps in tooling, architecture, and implementation quality.
• Incorporate peer technical reviews to validate findings and uncover blind spots that individual assessments may miss.
• Track DevOps Research and Assessment (DORA) metrics (e.g., lead time, failure rates) and treat declines in delivery or stability as signals to investigate through a training lens.

Core technical competencies to prioritize in the AI Era

Not every engineer needs deep machine learning expertise. Instead, organizations should distinguish between AI literacy (required for all engineers) and AI specialization (required for specific roles).

Competencies should be layered: foundational knowledge for everyone, applied skills for those building or integrating systems, and deployment-level expertise for teams running AI in production.

The table below maps four priority areas to relevant frameworks and role types:

AI and machine learning fundamentals

Every engineer working with AI-assisted tooling needs baseline AI literacy on how models are trained, where they fail, and how to evaluate outputs critically. Article 4 of the EU AI Act requires providers and deployers of AI systems to take measures ensuring a sufficient level of AI literacy among their staff, accounting for technical knowledge, experience, education, and the context in which the systems will be used. The obligation took effect in February 2025.

For enterprises operating in or selling into the EU, foundational AI training is a compliance requirement, not a discretionary investment.

Data engineering and cloud architecture

AI systems at scale depend on reliable data pipelines, well-designed lakehouse architectures, and distributed processing. Engineers who can’t build or evaluate these foundations create bottlenecks between model development and production deployment. Data engineering-focused learning content, like the resources on O’Reilly, should cover ETL/ELT design, data quality governance, and cloud native infrastructure at a minimum.

Cybersecurity and compliance

AI increases code velocity, which increases the surface area for security risk. NIST SP 800-218 includes “Training, Education, and Awareness” as a defined practice within its secure software development framework, grounding security training as part of the engineering control system rather than a downstream HR function. Role-based paths matter here since operational technology engineers and software engineers face different threat models.

Systems integration and automation

Deploying AI models into production requires mature CI/CD pipelines, container orchestration, and infrastructure-as-code practices. Toolchains such as GitLab and Kubernetes, paired with Terraform for infrastructure provisioning, are the practical environment where AI model iteration happens. Hands-on labs and sandbox environments accelerate skill transfer in this area more effectively than lecture-based formats.

What are the essential soft skills and mindsets engineers need in the AI era?

Training ROI isn’t determined by technical know-how alone. These four soft skill areas amplify the value of technical training across the organization:

• Cross-functional collaboration: AI projects require engineers, data, legal, and product teams to work together. Communication skills reduce costly handoff failures.
• Agile thinking and adaptability: AI tooling evolves quickly. Engineers who treat learning as continuous outperform those who treat each training as a final answer.
• Ethical decision-making: Understanding bias, fairness, and where models fail is a core engineering responsibility that affects deployment decisions and regulatory exposure.
• Continuous learning culture: Individual mindset shifts require organizational reinforcement through time, incentives, and visible leadership support.

Evaluating AI training formats for engineers: Online courses, certificates, workshops and more

Matching format to goal is as important as selecting the right content. Budget, timeline, and team cohesion requirements all shape which format delivers the best return.

On-demand online learning platforms

Flexible scheduling and lower per-seat costs make on-demand learning attractive for large engineering organizations. The practical risk is low completion without accountability structures built in, such as manager check-ins, cohort learning groups, or milestone-based incentives.

Accredited certificate programs

Certificates aligned to recognized standards, such as those grounded in NIST frameworks or ABET-accredited programs, carry external credibility useful for compliance justification and talent retention. O’Reilly’s certification preparation paths cover AWS, Azure, and security credentials.

Corporate workshops and seminars

Custom workshops can address specific technology stacks and team workflows that off-the-shelf content don’t reach. When evaluating vendors, prioritize instructor credentials, hands-on-versus-lecture ratio, posttraining support, and verifiable client references.

Success stories: How leading enterprises upskilled engineers for AI

Organizations across industries have moved from ad hoc upskilling to structured programs with measurable engineering outcomes.

HSBC and the Alan Turing Institute (banking)

HSBC committed to a multiyear, multimillion-pound partnership with the Alan Turing Institute to build internal data science capability. HSBC funded Turing researchers to work alongside its engineers on economic data science problems, embedding expertise through direct practitioner collaboration.

Open GI and Thoughtworks (software)

When Open GI migrated a flagship product to an Azure-based SaaS offering, developer upskilling happened in parallel with architectural transformation. Thoughtworks embedded Accelerate’s four delivery metrics into the team’s learning culture from the start. Lead time, deployment frequency, mean time to restore, and change fail percentage gave engineers a live feedback mechanism for their own skill development, while shared accountability across engineering, operations, and product replaced siloed ownership.

Target (retail)

Target spent over a year rebuilding its engineering culture through a DevOps transformation inside a large traditional IT organization. Engineering leads treated DevOps as an engineering capability to build through deliberate skills development, with tooling decisions following from that foundation. Lessons learned from each experiment were shared across the organization as they concluded.

Prepare your engineering teams for the next wave of AI innovation

The strategic roadmap for enterprise engineer training in the AI era covers three steps:

• Audit current skills against future-ready competencies using recognized frameworks like NIST NICE.
• Design layered learning paths that match role requirements, format constraints, and business timelines.
• Measure outcomes using the Kirkpatrick Model’s four levels, connecting behavioral change on the job to engineering delivery and security metrics.

O’Reilly supports teams working through exactly this kind of structured, role-based technical development, from the latest books and live courses on AI fundamentals to hands-on interactive labs covering cloud infrastructure, security, and DevOps.

Start a free trial to see how it fits your organization’s engineering training strategy.

FAQ