From Sysadmin to AI Solutions Engineer: A One-Year Learning Roadmap

From sysadmin to solutions engineer: why I’m spending the next year learning AI in production

Sysadmin Jay Thomason is leveraging his role at a law firm to spend one year transitioning into AI solutions engineering. He manages a production environment with 85 users across three offices, providing a real-world lab for testing AI implementations.

Why This Matters

Moving from theoretical AI tutorials to production environments reveals the friction of legacy systems and user requirements. Thomason argues that “vibe-coding” with tools like Claude is acceptable as long as the engineer can debug and explain the underlying architecture, a necessary shift for IT professionals facing the rapid evolution of agentic workflows and RAG patterns.

Key Insights

• Prisma Health domain migration for 25,000+ users (2020) serves as foundational experience for large-scale enterprise transitions.
• Retrieval-Augmented Generation (RAG) is defined as the workhorse pattern for modern AI, requiring specific expertise in chunking and embedding strategies.
• The AB-410 Intelligent Applications Builder certification is replacing the PL-200 in November 2026 for Microsoft partner ecosystems.
• Azure OpenAI and Microsoft Graph API can be combined to automate offboarding runbooks and prevent checklist drift in production IT shops.

Practical Applications

• Use case: Automating law firm offboarding via Microsoft Graph API to synchronize live state with licensing. Pitfall: Skipping hands-on lab exercises in certification paths leads to wasted time during implementation.
• Use case: Developing internal RAG knowledge bots for legal documentation retrieval. Pitfall: Over-reliance on AI-generated code without the ability to debug or extend it independently.

References:

• https://dev.to/jaythomason/from-sysadmin-to-solutions-engineer-why-im-spending-the-next-year-learning-ai-in-production-59d3