The gap between what you were promised and what you're being asked to do is real, and it's not your fault. You were hired into a technical role, and instead you're being funneled into data governance and product management, work that holds no interest for you and offers no path toward the hands-on model building you want. This is a common story at large consultancies, where junior talent is often treated as a flexible resource to fill whatever client need arises, regardless of the employee's stated career direction. The firm's priority is utilization, not your technical growth. That mismatch is costing you time and exposure you can't get back.
The practical question is whether you can pivot in six to twelve months. The answer is yes, but only if you stop waiting for your current employer to give you the right projects. You already understand the theory behind AI and have experimented in sandboxes. That is a stronger foundation than you may realize. Many practitioners started exactly where you are: with conceptual knowledge and a willingness to build. What you lack is production experience, and that is the single most important credential for the roles you want. You need a demonstrable project, something you can point to and say, "I built this, I validated it, I pushed it live." That project does not have to come from your day job. It can come from a personal initiative, an open-source contribution, or a freelance engagement. The key is that it must be real, end-to-end, and recent.
Your analytical background gives you a running start. You already know how to think critically about data. Now you need to close the gap on deployment: containerization, CI/CD pipelines, monitoring, and model serving. These are learnable skills, and the tools are more accessible than ever. Spend your evenings and weekends building. Document your process. Share it publicly. When you interview, you will not be selling your current job title, you will be selling the work you chose to do. Employers in technical roles care about what you can ship, not about the governance meetings you were forced to attend.
One more thing: be direct about what you want. In your next role, whether at a different firm or a smaller team, ask specific questions about the ratio of hands-on technical work to strategy or management tasks. If the answer is vague, walk away. You don't have to become a data governance person. You just have to stop letting someone else decide your path for you.