Executive Operations · Workflow Improvement · Responsible AI

Turn operational friction into tested, usable workflows.

I work where operations, partners, data and digital tools meet.

My background is hands-on operations, purchasing, planning, ERP data and cross-functional follow-through. I use that operating discipline to clarify requirements, test AI-assisted outputs against trusted information and real exceptions, and keep human ownership visible.

Greater Copenhagen · French native · English fluent · Danish operational (PD2/B1+; SAP S/4HANA course completed fully in Danish) · EU citizen

At a glance

What I bring to an AI-enabled operations team.

Operational leverage for a key manager: project and administrative follow-up, partner coordination, reliable information and the business-side work that makes an AI-enabled workflow useful, testable and accountable.

01

Execution & follow-through

Turn decisions and agreements into visible actions, owners, dates and escalation points, reducing the follow-up load around a key manager and staying with the issue through closure.

02

Partners & stakeholders

Coordinate external partners and internal teams without losing the operational consequence, commitment or next action.

03

Data & process control

Question source quality, reconcile conflicting information and make downstream impact visible before a decision is released.

04

AI workflow testing

Translate a recurring business problem into requirements, test outputs against real and unusual cases, document failure and retest.

Operating method

I do not start with AI. I start with the problem.

AI is useful when the source, owner, decision boundary and failure behaviour are clear enough to test.

01

Friction

What is repeatedly unclear, slow, manual or error-prone?

02

Source

Which information is trusted, current and allowed to be used?

03

Owner

Who decides, who acts and who must review exceptions?

04

Build

Where can automation or AI remove repetitive work without hiding judgement?

05

Test

Run normal cases, edge cases, missing inputs and contradictory inputs.

06

Review

Compare the output with operational reality and the expected decision boundary.

07

Improve

Log the issue, explain the failure mode, adjust requirements and retest.

08

Handover

Leave clear ownership, limits, user guidance and a repeatable review rhythm.

Business-side AI qualification

How I would test an AI agent before trusting the workflow.

The objective is not to make the agent look clever. It is to find where it fails before the failure reaches a customer, partner, commitment or decision.

01

Define expected behaviour

What should the workflow produce, from which approved sources, and what must never be inferred?

02

Build realistic test cases

Use routine examples, incomplete information, conflicting information and unusual but valid situations.

03

Challenge the output

Compare the result with source data, operating reality, policy and the decision boundary. Record why an output is wrong, incomplete or unsafe.

04

Feed back, retest, hand over

Translate findings into clear issues or requirements for the technical team, retest the changed behaviour and document the remaining limits for users.

TESTINPUTEXPECTED BEHAVIOURREVIEW
Normal caseComplete approved source dataCorrect structure and expected next actionBusiness user
Missing inputOne required source absentAsk / stop instead of inventingBusiness user
Conflicting inputTwo sources disagreeFlag conflict and identify decision ownerProcess owner
Edge caseUnusual but valid situationStay inside rules and surface uncertaintyProcess owner
High-impact exceptionOutput could affect customer, partner, cost or commitmentHuman review before actionResponsible manager

Illustrative validation pattern. It demonstrates how I would structure business-side testing; it is not a claim of ownership of a production AI platform.

Transferable evidence

The judgement behind the workflow comes from real operations.

I do not present AI as a substitute for operating experience. These cases show why I can recognise bad data, unclear ownership, downstream impact and incomplete handovers.

01DATA QUALITY + SYSTEMS

Trusted by IT to clean source data before it became a decision problem.

At A.K.S. Finmekanik, the IT Manager asked me directly to clean duplicate, obsolete, incorrect and ghost/orphaned records. During incoming customer PO/order and related commercial checks, I also identified conflicting cost/sales-price data affecting 100+ items, traced the issue and coordinated correction. The transferable point is simple: a system output is not automatically a reliable answer.

Evidence: IT-requested cleanup · 2,000+ records · 100+ affected items · source tracing and correction
02EXECUTION

Kept supplier commitments and operational consequences visible under pressure.

Across 30+ international suppliers, I worked with confirmations, shortages, lead times, alternative sourcing, delivery recovery and cross-functional escalation. When increased volume constrained a capable supplier, I controlled commitments and redirected urgent demand instead of waiting for the problem to resolve itself.

Evidence: supplier follow-up · prioritisation · recovery actions · owners and next steps
03CHANGE + HANDOVER

Translated changing requirements into controlled operational actions.

Daily AX/D365 work included 2,000+ ERP/master-data records and 200+ complex product structures. Changes could affect items, BOMs, routings, material need, supplier timing, quality documentation and production readiness, so impact, validation and handover had to stay connected.

Evidence: ERP · planning · Engineering · Quality · Production · suppliers · documentation
04LIVE IMPLEMENTATION

Turned unclear owner input into a real customer-facing operating system.

For Marzieh Nail Atelier, I helped structure services, pricing, policies, booking paths, multilingual information, QA, release and owner handover. It is a small-business implementation, not an enterprise transformation claim, but it proves that I can move from ambiguity to something real, tested and maintainable.

Evidence: requirements · source-of-truth structure · EN/DA · QA · release · handoverView live implementation

Business-to-development bridge

The best use of my profile is between the daily operation and the people building the tool.

MY CONTRIBUTION

Operational translation and validation

  • Clarify the recurring problem and desired outcome
  • Identify trusted sources, owners, exceptions and review points
  • Write practical requirements and realistic test scenarios
  • Test outputs against real operating situations
  • Document failures, questions and user feedback clearly
  • Retest and help prepare user guidance and handover
TECHNICAL TEAM

Engineering and deployment

  • Agent and model engineering
  • Platform architecture and infrastructure
  • System integrations and technical security controls
  • Production deployment and monitoring
  • Technical remediation when tests expose a failure
  • Versioning and platform-level release management

The bridge is the value: turn operational reality into something developers can build and users can trust.

Applied workflow practice

Practical use, with the boundaries kept visible.

My recent independent work has focused on how operational information can be made clearer, easier to review and safer to reuse with modern tools.

01

Workflow analysis

Map the current process, repeated friction, inputs, exceptions, owners and handoffs before proposing a tool.

02

AI-assisted work

Use AI for structured research, comparison, drafting, documentation and prototype support while keeping source boundaries and human approval explicit.

03

Testing discipline

Define expected behaviour, use test examples, challenge unexpected output, record issues and retest after changes.

04

Privacy & control

Keep confidential, personal, export-controlled or restricted information inside approved systems and processes.

WHAT I CLAIM

Business-side workflow analysis, source discipline, requirements, practical testing, documentation, implementation support and human-review logic.

WHAT I DO NOT CLAIM

AI model engineering, platform architecture, autonomous deployment, or professional ownership of a specific employer's internal AI stack.

Best-fit roles

Where this combination is useful.

01

Executive Operations

Decision and meeting follow-up, practical administrative support, action control, progress visibility and cross-functional execution close to a key manager.

02

Partner Operations

External-partner coordination, commitments, issue resolution and operational communication.

03

AI Adoption Support

Business requirements, trusted sources, test scenarios, output QA, feedback and human-review boundaries.

04

Project Coordination

Owners, dependencies, risks, actions, progress and practical handover.

05

Process Improvement

Find recurring friction, clarify the source of truth and create a more repeatable operating rhythm.

06

Business Systems

ERP/master data, operational reporting, workflow documentation and business-to-technical translation.

Contact

Looking for someone who can keep the operation moving while making the workflow smarter?

I bring the operational discipline first, then use data, process design and AI-assisted tools where they make the work clearer and more reliable.

Email RomaricDownload CVTune, Greater Copenhagen, Denmark
clementiromaric@protonmail.com
+45 28 73 74 85