OpenAI · Forward Deployed Engineer (Zurich)

LinkedIn

Enterprise AI leader, builder and founder.

Turning frontier AI into production across startups and large organisations. AI Lead at Zurich Airport, founder of KI-Unlocked, helping 10+ companies become AI-native. GPT-3 since 2020, with 15+ AI applications built.

Zurich Airport
UBS
PwC
World Economic Forum
SAP
Forbes
TEDx
HSG

Why me

What I bring to OpenAI on day one.

Enterprise deployment

Discovery to production to governance. Full path.

Discover
Build
Deploy
Adopt
Govern

AI builder

15+ products. Agents, RAG, voice AI, eval loops.

Business x Engineering

Translate between executives, engineers and end users to keep projects moving.

BusinessEngineeringUser

90-second walkthrough

Three recent deployments

Zurich Airport - Voice AI Agent

Voice AI for 85K annual calls. Discovery through formal tender.

~85K

calls/year

8 FTE

call center

Regulated

environment

My role

Discovery · Use-case prioritisation · Architecture and prototype · Vendor evaluation · Governance

Outcome

PoC in two weeks. Formal tender initiated. Procurement ongoing.

ZRH Insider - Employee App Rebuild

Three codebases merged into one foundation for 35K employees.

35K

employees

3 to 1

codebases merged

My role

Rebuilt app foundation in one month · Consolidated architecture for faster rollout · Designed engagement loops

Outcome

Shipped. Externally quoted at CHF 350K. Scalable base for future features.

Engineering Office LAJO - AI-Native Operations

Company knowledge base powering proposals and project intelligence.

98%

accuracy rate

60%

time saved per proposal

My role

PoC ready before first sales meeting · OCR + model routing + RAG pipeline · Secured paid workshop, now in delivery

Outcome

Won trust before the first meeting. Knowledge base live. Proposal automation next.

How I move from zero to deployed

Five stages. Pick one to see the layer, what runs, and what blocks it.

DiscoverDecideBuildDeployScale

Discover

Layer: Intake

Map stakeholders, data boundaries and who can say yes.

What runs

Call arrives, language detect, intent classify

What blocks it

Procurement cycles and too many stakeholders, no single owner.

Selected projects I created from scratch this year

Personal products. AI infrastructure, voice, human intelligence.

01Live
✓
Context protected
Identity hidden OnRaw data stored Off
AI infrastructure

PrivacyLayer

Browser-side PII stripping before prompts reach any model. Zero bytes to any server.

EvidenceLive product with Chrome extension
View project
02In delivery
Project intelligence
Project A12DecisionsExpertsStandards
Why did we change the load assumption?3 cited sources
Enterprise AI

Engineering Office OS

RAG + OCR pipeline connecting project history, decisions and expertise across offices.

EvidenceVertical system designed
03Building
Human intelligence

Vibe Translator

Context layer decoding tone, personality and culture before generating responses.

EvidenceInteractive product prototype
View project
04Venture
ZRHHKG
Next: Gate E46

12 min walk · passport control ahead

Travel

AirCompanion

Contextual AI companion for the full airport journey. Real-time, location-aware.

EvidenceProduct system designed
View project
05Concept
“Let's consider it.”
UNSPOKEN CONTEXTThis may be a polite “no.”
Cultural intelligence

Asia Bridge

Communication rules and social nuances of Asian cultures. Duolingo meets Grammarly.

EvidenceProduct thesis developed
View project
06Live
AI
Procurement negotiation
Simulation

Rehearse

AI role-play for sales, negotiations and difficult conversations. Real-time feedback.

EvidenceInteractive rehearsal flow
View project
07Testing
Mum's story · 04:28

“I never told you how we first arrived…”

Voice AI

Treasure

Voice AI that captures family stories and preserves them across generations.

EvidenceVoice-first experience built
View project

I'd rather show you than tell you.

30 minutes. I'll walk you through a live deployment, the decisions behind it, and where I think OpenAI's enterprise play should go next.