Blue/Green Testing My AI-Enhanced CV: A Data-Driven Approach to Job Applications
(Note for US readers: CV, or Curriculum Vitae, is the standard term in the UK and many other countries for what Americans call a resume. While traditionally a CV might be longer and more detailed than a resume, the terms are often used interchangeably in today's international job market.)
Introduction
As a network architect who's spent decades building resilient systems, I recently found myself applying engineering principles to an unexpected domain: my CV. What started as a simple experiment with AI tools evolved into a fascinating exercise in A/B testing and data-driven optimisation. The project has not only yielded interesting initial results but has also provided insights into how AI might transform the way we present ourselves professionally.
The Genesis of the Experiment
Like many professionals, I initially approached AI-assisted CV writing with scepticism. However, as someone who regularly works with emerging technologies, I decided to treat it as a technical challenge rather than a magical solution. This led me to develop a systematic approach to feeding the AI with comprehensive data about my professional experience and target roles.
The first step was building a robust knowledge base. I gathered multiple versions of my CV spanning the past decade, including role-specific variations and comprehensive work histories. I collected current job descriptions from my target market, spanning cloud network architect roles, technical design authority positions, and platform engineering leadership opportunities. To provide additional context and demonstrate expertise, I included my published technical articles from Medium, white papers on cloud architecture, and anonymised non-sensiive documentation from older projects.
The Methodology
The process of working with AI to generate CVs proved more complex than simply feeding in data and receiving a polished output. I developed a staged approach that allowed for better control and refinement of the results.
The first stage focused on understanding each role. I would feed the AI a specific job description and have it analyse the key requirements and desired experience. This helped identify areas of my experience that needed emphasis and any potential gaps that required attention.
Content generation formed the second stage, where I would request role-specific versions focusing on relevant experience. The AI would generate various ways to present key achievements and technical project descriptions, providing multiple options for consideration.
The refinement stage proved crucial. Here, I would compare AI-generated content with my original materials, identifying any misunderstandings or over-emphasis. This stage required particular attention to maintaining technical accuracy and ensuring the language and tone remained appropriate for the UK market.
The Hyperbole Challenge
One of the most significant challenges emerged early in the process. The initial versions read like they were written by an overenthusiastic recruiter after three espressos. While technically accurate, they were laden with superlatives and grandiose statements that felt out of place in the measured tone of UK technical circles.
Common infrastructure upgrades were described as "revolutionary transformations." Standard implementations became "groundbreaking innovations." My cloud networking experience was invariably "unparalleled." This Silicon Valley-style self-promotion required significant toning down to match the more reserved expectations of the UK market.
The Blue/Green Testing Framework
Drawing on my experience with infrastructure deployments, I implemented a blue/green testing approach to measure the effectiveness of different CV versions. The traditional, manually updated CV serves as the "blue" version, while the AI-enhanced, role-optimised approach represents the "green" version.
The testing process involves alternating between versions for similar roles and tracking responses systematically. I'm documenting response rates, time to first response, and the quality of role matches. However, it's important to note that the sample size is currently too small to draw definitive conclusions about which approach is more effective.
This systematic approach to testing has already provided valuable insights into how different organisations respond to varying presentations of the same professional experience. It's also highlighted the importance of controlling for variables beyond the CV itself, such as seasonal variations in hiring and changes in the job market.
The Changing Landscape of Professional Presentation
This experiment takes place against the backdrop of an industry in transition. The traditional CV writing and review industry, which has long charged significant fees for professional services, faces disruption from AI tools that democratise much of their expertise.
Rather than disappearing, these services are likely to evolve. Some CV writers are already incorporating AI tools into their workflows, using them for initial drafts while focusing their expertise on refinement and personalisation. Others are shifting their focus to coaching clients on effectively using AI tools for CV creation and optimisation.
The widespread availability of AI tools presents new challenges for both job seekers and recruiters. As AI-generated CVs become more common, standing out authentically becomes increasingly crucial. Recruiters and hiring managers are developing new strategies to identify genuine experience and capabilities, while the focus may shift from the CV itself to supporting evidence of expertise.
Looking Forward
While my testing continues and the sample size grows, several things have become clear. AI can be a powerful tool for optimising how we present our professional experience, but it requires careful management and human oversight. The key lies in finding the right balance between leveraging AI capabilities while maintaining authenticity and personal connection.
The future of professional presentation likely lies in this blend of human expertise and AI assistance. The most successful approach will combine AI's ability to identify patterns and optimise content with human judgment about how best to present our professional stories.
For now, I continue to refine my methodology and gather data. The goal isn't to create a perfect CV but to develop a systematic way of presenting experience that resonates with different audiences while remaining authentic. Like any good engineering solution, it requires monitoring, measurement, and continuous improvement.
The CV writing industry itself, and much of recruiting in general, stands at a crossroads. Success will likely come from embracing AI as a complementary tool rather than viewing it as a threat. The focus may shift from pure writing services to helping clients navigate the increasingly complex world of professional presentation in an AI-enabled world.
In the end, this experiment has reinforced something I've long believed about technology: it's most effective when it enhances rather than replaces human capabilities; AI is a tool like any other that has come along in the last few decades of my career and maintaining a balance of human interaction will be crucial for authentic professional presentation.