Portfolio kompetencji Marcina Zycha dla stanowiska AI Implementation Specialist. I integrate AI models (GPT-4, Claude, Gemini) into business processes. From concept through MVP to production. 16 years of experience at the intersection of technology and business. Kluczowe wyniki: 104 completed projects, 16+ years of experience, 1M+ customers in systems.

Position

Most AI deployments end at the prototype stage.

AI that earns — not AI that impresses on slides

Most AI projects die between proof of concept and production. I deploy AI models that work in real environments — with your company's data, in your processes, with measurable ROI.

104 completed projects
16+ years of experience
1M+ customers in systems
References available on request. Certifications: IAB DIMAQ Professional, Google Analytics Advanced, IT Project Manager (Laba), Business Analyst (Laba).
The value I bring

What sets me apart

7 days MagnAI payback

Business First

I start with the business problem, not the technology. AI is a tool, not an end in itself.

ROI analysis before the project, not after. Business metrics — not just technical ones.
104 completed projects

End-to-end Ownership

From discovery through implementation to adoption.

104 projects taken to production, not prototypes
50+ MarTech projects

Pragmatic Tech Choices

I choose the simplest solution that works. Sometimes that's AI, sometimes a regex, sometimes a human.

Off-the-shelf APIs instead of custom ML, proven solutions instead of experiments
104 projects in production

Change Management Included

Technology without adoption is failure. Training, documentation and support are part of every project.

ADKAR framework, adoption metrics, post-deployment support

I don't sell technology. I solve business problems.



Detailed portfolio available in CV

Full project history, certifications and references

View full CV

My AI implementation methodology

A repeatable process from discovery to production

Phase Time Timeline (weeks)
1 Discovery
1-2 wks
2 Design
1-2 wks
3 Prototype
2-3 wks
4 Build
4-6 wks
5 Deploy
1-2 wks
6 Optimise
ongoing
MVP Ready Production
0 2 4 6 8 10 12 14 wks
Typical time to production: 10-14 weeks (with working prototype after 4-5 weeks)

Detailed skills analysis

Skills breakdown across 5 key categories

AI & Machine Learning 85%
LLM Integration Prompt Engineering API Development
Data Engineering 90%
BigQuery SQL ETL Pipelines Data Modelling Power BI
Project Management 90%
Agile/Scrum Stakeholder Management Risk Analysis Resource Planning Documentation
Change Management 80%
ADKAR Framework Training Design Adoption Metrics Communication Strategy
Business Acumen 95%
ROI Analysis C-level Communication Budget Management Strategic Planning KPI Design
Overall match score 87%

MagnAI — AI Content Automation System

Manual product description creation — too slow and too costly

6 weeks ROI: 7 days
Metric Before After Change
Descriptions/day manual 1000+ CV: MagnAI
Cost reduction copywriter -95% CV: MagnAI
ROI payback 7 days CV: MagnAI
Tech stack:
GPT-4 Claude API Python FastAPI React

Key takeaways

Prompt engineering more effective than model training Human review critical in the first 2 weeks Quality metrics matter more than speed

Change management

Technology without adoption is a cost, not an investment. Every AI implementation is an organisational change project.

ADKAR Framework Change management model
100% Adoption rate
~10 wks Full cycle
Wk. 1-2
A

Awareness

Awareness

Wk. 2-4
D

Desire

Desire

Wk. 4-8
K

Knowledge

Knowledge

Wk. 6-10
A

Ability

Ability

Wk. Ciągłe
R

Reinforcement

Reinforcement

Awareness
"Why is the change needed?"
  • Executive presentation with business case
  • AI capabilities demo on real company data
  • Competitor and market trend analysis
Desire
"What's in it for me?"
  • "A day with AI" workshops for teams
  • Early adopter identification
  • Quick wins in the first 2 weeks
Knowledge
"How does it work and how do I use it?"
  • Role-specific training
  • Documentation with examples
  • Office hours and Q&A sessions
Ability
"Can I do this independently?"
  • Supervised practice with feedback
  • Checklists and ready prompts
  • Peer mentoring programme
Reinforcement
"How to sustain new habits?"
  • Adoption metrics in dashboard
  • Recognition programme for power users
  • Iterative improvements based on feedback

Anti-patterns to avoid

Mistakes that torpedo AI implementations
Big-bang deployment without a pilot
"One-off" training with no follow-up
No executive sponsor
Ignoring team resistance and concerns

Relevant experience

Roles and projects related to this area

2024 - 2025

ETOS S.A. / Diverse

Performance Marketing Coordinator

Deployed LTV system for 1M+ customers with ECDP integration. Team ROI 2308%, payback 13 days. Budget: 6.5M PLN.

  • LTV system for 1M+ customers
  • Team ROI 2308%, payback 13 days
  • Budget 6.5M PLN
2018 - present

Marcin Zych

AI & MarTech Consultant

50+ implementation projects. GPT-4, Claude, Gemini API integrations. Author of MagnAI — AI automation system.

  • 50+ implementation projects
  • GPT-4, Claude, Gemini integrations
  • Author of MagnAI
2022 - 2023

Yetiz Interactive

Senior Facebook Ads Specialist

Built Power BI analytics models for 15+ clients. ROAS optimisation +40%.

  • Power BI for 15+ clients
  • ROAS optimisation +40%
2019 - 2022

OMNIOXY S.A.

Performance Manager / Technical Marketing Manager

Implemented GTM and Data Layer architecture. Landing pages with CRO. Performance marketing management, budget 2M PLN.

  • GTM & Data Layer architecture
  • Performance marketing + CRO
  • Budget 2M PLN
16+ years of experience
104 completed projects
1M+ customers in systems

Key projects

Case studies with measurable business impact

AI/ML

MagnAI

95% cost reduction vs copywriter, payback in 7 days

-95% cost
7 days payback
Data Pipeline

LTV System

1M+ customers, RFM segmentation, churn prediction

1M+ customers
2308% team ROI
Integration

Server-side Tracking

Data quality increased from 70% to 99%

99% data quality
+29pp improvement
Analytics

Cross-channel Attribution

Budget misallocation reduced by 40%

-40% misallocation
+40% ROAS
Analytics

Power BI Dashboards

Automation from 8h/week to 15 min/day

15 min per day
-97% time
Automation

AI Description Generator

1000+ product descriptions/day, 95% cost reduction

1000+ desc/day
-95% cost
Discovery Framework

Questions at the discovery stage

Key topics I cover in the first meeting

Before every implementation I conduct a detailed interview to understand the business context and define measurable goals.

01

Business Context

4 questions
What business problem do you want to solve with AI?
Which KPIs do you want to improve and by how much?
Who is the primary beneficiary of this solution?
What is the deadline and what drives it?
02

Current State

4 questions
What does the current process look like (AS-IS)?
How much time/money does the current approach cost?
What tools are already in use?
Have there been previous automation attempts?
03

Constraints

4 questions
What are the data security requirements (GDPR, industry-specific)?
Are there budget or time constraints?
Which systems must be integrated?
Who needs to approve the deployment?
04

Success Criteria

4 questions
How will we know the deployment succeeded?
What metrics will we measure?
What would be a "home run" vs. minimum viable?
What is the fallback plan if AI doesn't work?

Interested in working together?

Book a free 30-min consultation — a concrete business problem, a concrete answer, no commitment.

Free 30-min consultation

A concrete business problem, a concrete answer. No commitment.

Book a 30-min consultation Download CV (PDF)
Open to opportunities

LinkedIn

Full career history, recommendations and professional activity.

LinkedIn
Available now B2B or Employment Warsaw / Remote