Welcome to Gradient Drift!
This is my blog chronicling my journey over the next 9 months as I battle to gain AWS’s brand-spanking new Generative AI Developer Professional Certification.
Over the last year I have gained most of AWS’s Associate-level certs spanning my core interest areas, cloud architecture, data engineering, machine learning and AI. However I recently learned that AWS is retiring the Machine Learning Specialty cert, which had previously been next on my bucket list - to be replaced by this new certification, which is abbreviated to GAD-C01.
This cert has really captured my imagination in recent days. I had previously been underwhelmed about the practical value of Generative AI that currently exists today. While ChatGPT and Claude are obviously amazing (insofar as I practically view them as immediate family-members at this point), my experimentation in building applications consuming leading foundation models for inference, have so far produced pretty lacklustre results (to say the least..). My view until now had been that there is far more value in traditional machine learning, and I still hold that view. However I now believe that it is foolish to view GenAI based on its capabilities today. The frontier models of 2026 and beyond are likely to be very different entities, with far-reaching societal implications as widespread adoption innevitably accelerates next year.
For that reason I feel that it is imperative to gain this certification. Firstly the rate at which AWS has stepped up its marketing of the 4 certs I gained this year in recents weeks, suggests that their value is likely to erode quite soon as more people jump on the AI bandwagon. Associate level certs are no longer sufficient, I have to up my game immediately. Its also quite likely that the lay-off waves synonymous with tech, will rapidly proliferate into many other sectors of the UK economy over the next year. And increasingly the value of employees will be based on their ability to leverage AI Native tools to exponentially increase their output. While this isn’t really a thing today, this will change very soon.
What I love most about this certification is that it is centred on development, i.e. the hands-on delivery of production-grade applications. I am far from a developer, but I am definitely someone looking to put tutorial purgatory firmly behind them once and for good, and so over the course of the next 9 months leading to me taking the exam, I will be putting together numerous projects leveraging key AWS GenAI services (e.g. AWS Bedrock, API Gateway, Amazon Guardrails for Bedrock, Step Functions, S3, Lambda & Kendra). Key themes will include Foundation Model Selection (i.e. Titan, Claude and Llama et al), Retrieval Augmented Generation (RAG), Guardrails, Agentic Orchestration, Vector Databases and Prompt Engineering.
Anyway it is my intention to share regular blog posts outlining the projects I build from inception to final delivery and all the lessons learned along the way. Thank you for your patience thus far. Hopefully future posts will become much easier to read, as I slowly regain my ability to write by reversing this tendency I've developed where I instinctively submit my editorial autonomy over to our dear friend ChatGPT. In any case, my next post will contain a roadmap of the projects I'll be buiding, as well as the rationale behind them, between now and exam day (circa July 2026).