whoami

Enterprise AI
doesn't have to live
in the cloud.

I'm Ali Ozdemir — a Solution Architect working at the seam between large-scale Salesforce delivery and local LLM infrastructure. I write about what happens when regulated enterprises stop sending their data to someone else's GPUs.

~/now
updated may 27
currently // co-leading the Salesforce CoE for DACH in valantic, working on CRM, AI & Data Cloud rollouts.building // a Golang toolkit for running quantized models inside enterprise networks and various productivity tools.reading // AI progression from various LLM companies.tinkering // Raspberry Pi Home Server, ESP32 + on-device whisper for an offline voice agent.

Notes from the seam.

Long-form thinking on enterprise architecture, local AI in regulated environments, and the unfashionable parts of solution design.

The walled garden ends at integration — a Salesforce architect's distributed-systems guideWhy I added Go to my Salesforce toolkit - and what broke firstWhy I stopped sending PII to OpenAI — and what I ran instead

Major Things I've shipped.

A few projects that taught me more than the rest. Each one with the architecture decisions, the tradeoffs, and what I'd do differently.

Global Unified Loyalty & AI Agent Program

2024 — present
valantic dmc · global retail

Orchestrated a multi-cloud transformation across 17 countries — Service Cloud, Loyalty Management, Marketing Cloud, Data Cloud, and Agentforce stitched into one event-driven loyalty platform.

17
countries
500k+
daily events
6
product teams aligned
30+ Mio
total customers
read the case study

First-to-Market Agentforce in DACH

2025
salesforce strategic partner

Architected and delivered one of the first Agentforce implementations in Germany. The part nobody talks about: getting LLM-powered features through an enterprise security review.

1st
in DACH
4 wk
from kickoff to live
read the case study

Transform SAP IT Landscape to Salesforce

2023 — 2024
german enterprise operating globally · b2c

Solution design for complete CRM and MarTech transformation from SAP to Salesforce — CRM migration, marketing transformation and loyalty program design.

12
countries onboarded
5
systems migrated
8
months of time
read the case study

Global CRM and Data Platform Rollout

2020 — 2022
french enterprise · b2c

Orchestrated a multi-cloud transformation across 3 continents — eCommerce, Salesforce Core, Marketing Cloud, and Data Cloud forged into a data powerhouse.

3
continents
200+
daily users
read the case study

Omnichannel Commerce Scale-up

2022 — 2023
us enterprise · b2c

Took a separate subsidiary landscape and turned it into a single enterprise program — CRM migration, subsidiary onboarding, the lot.

8
subsidiaries onboarded
SAP↔SF
integration backbone
read the case study

The workshop.

Where the ideas in the writing actually get tested. Most of these live on GitHub. Some of them work.

A few facts.

I've been a Solution Architect for eight years, mostly inside the Salesforce ecosystem, mostly on the kind of multi-cloud rollouts where the org chart is more complex than the data model.

What I've gotten interested in lately is the part where that world meets the local-AI world: regulated enterprises that want generative features but can't legally put their data through a third-party API. The interesting work is in the seam.

Current focus is Hardware & IoT, Golang and CLI apps to build arms and legs for the local-AI stack and boost productivity via the toolkit I am building.

After software I studied bioengineering MSc, which is an oddly useful background for thinking about systems where the failure modes are the point. I also play a fair amount of piano.

Want to talk shop?

I'm open to talks, podcast guests, advisory work, and the occasional genuinely interesting consulting engagement.