AI Competence Center
One portfolio for AI initiatives that were running in six IT functions separately: delivery, architecture, business analysis, QA, development, DevOps.
00
AI ENABLEMENT
I built the AI Competence Center. AI work was running separately in six IT functions — delivery, architecture, business analysis, QA, development, DevOps. It’s one tracked portfolio now, from idea to pilot to production.
SCROLL
THE APPROACH
AI initiatives were running in six IT functions and nobody had a single view of them. I built the AI Competence Center to hold them in one place. Every initiative now passes the same intake template: purpose, assumptions, risks, business value. Then it moves on a tracked path — idea, pilot, production.
01
One intake, four questions.
Every initiative states its purpose, its assumptions, its risks and its business value before it gets built.
02
One path: idea, pilot, production.
Every initiative is tracked through the same three stages.
03
A written AI policy.
I wrote the AI policy my employer adopted, after working through the EU AI Act and GDPR.
SELECTED WORK
One portfolio for AI initiatives that were running in six IT functions separately: delivery, architecture, business analysis, QA, development, DevOps.
Connected AI agents to enterprise tools such as Jira and Confluence through MCP servers.
A library of 20 AI skills and around 50 prompts, shared across teams.
Role-based AI enablement for more than 50 employees, designed around what each role actually does.
AI functionality in the internal employee-discount portal. It picks the most relevant discounts for each employee’s weekly email. In pilot, company-wide.
WHAT CHANGED
6
IT functions, one portfolio
Current
20
AI skills, ~50 prompts
Current
2→1
Weeks to prep a proposal
Current
45–50
Proposals a month, from ~30
Current
50+
People enabled by role
Current
~$800K
AI deals contributed to
Current
~50%
Cut in call-centre staffing requirements
2023–2025
23%
Lower operating costs
2023–2025
EXPERIENCE
Feb 2025 – present
One progression, three titles.
Jul 2025 – present
Lead AI Specialist
Built the AI Competence Center and the standards every AI initiative passes through.
May – Jun 2025
Project Manager
Feb – Apr 2025
Python Developer
Delivered a 12-week scope in 8 weeks.
Jul 2023 – Feb 2025
Fertinagro AgroVip — fertiliser manufacturer, part of Grupo Tervalis
Automated the main call-centre workflows with a conversational bot.
Apr 2021 – Jun 2023
MonkeyLab
Coordinated a 15-person cross-functional team.
Sep 2021 – May 2024 · alongside the work above
Indiana University
Luddy School of Informatics, Computing, and Engineering.
SPEAKING
I speak at webinars and events as an expert on AI adoption, AI in the software development lifecycle and broader AI topics.
Most AI adoption stalls at the pilot. The fix is not a better model — it is an intake standard and a stage gate.
Marat Kornyev · Lead AI Specialist
AI in the software development lifecycle is not one tool per team. It is one portfolio across delivery, QA, development and DevOps.
Marat Kornyev · Lead AI Specialist
An agent is only as useful as the tools it can reach. That is what MCP is for.
Marat Kornyev · Lead AI Specialist
Governance is not the brake on AI adoption. Without it, nothing reaches production.
Marat Kornyev · Lead AI Specialist