Three kinds of work, one way of doing it. Jump to data from the web, AI, safely or your own data.
- You own what I build. Code, data and documentation are yours. You can take the system with you.
- It runs on infrastructure you control. Your own servers or a host you choose, and AWS only where it is the right tool.
- Costs you can predict. No open-ended cloud bills that grow with every request.
Data from the web
For recruiters, sales and lead generation, market research, real estate and price monitoring
Sounds familiar?
- We copy data from websites by hand, every week.
- We need every new job posting or listing in our market as soon as it appears.
- We tried a scraping tool, and it broke the moment the website changed.
What I do
- Custom scrapers and recurring data feeds: job postings, company data, listings and prices.
- Data that arrives cleaned, deduplicated and enriched, as an API, a database or plain files.
- Monitoring that keeps feeds running when websites change.
- Collection within the law: public data only, never anything behind a login.
Proof: Data Department, including its job postings data, and WPOptic. Writing: posts about scraping →
Tell me which websites and what data you need. Email erik@datadepartment.io, call +31 6 52 49 89 64 or send a message.
AI in your business, safely
For companies adopting AI, LLMs and agentic workflows
Sounds familiar?
- Staff paste customer data into AI chat tools, and nobody knows what is allowed.
- We think AI could save us time, but we don’t know where to start.
- Our data can’t leave the building, so cloud AI is out.
What I do
- An assessment of where AI actually saves time in your processes.
- LLM pipelines and agentic workflows: extracting data from documents, classifying, internal assistants.
- Clear rules on what data may go to which model, with self-hosted open models when data can’t leave the building.
- Access control, logging and protection against prompt injection.
- The GDPR/AVG side of AI use.
Proof: self-hosted LLMs on my own GPU hardware, pipelines that distill a large model into small ones, and fine-tuning small open models. Writing: posts about secure AI →
Tell me which task you would like AI to take over. Email erik@datadepartment.io, call +31 6 52 49 89 64 or send a message.
More from the data you already have
For webshops and other companies sitting on years of customer data
Sounds familiar?
- We have years of order data, but we only look at last month’s revenue.
- Customers buy once, and we never hear from them again.
- The related products in our webshop are picked by hand.
What I do
- Recommendations: product-to-product and based on what a customer bought before.
- Customer analysis: segments, RFM scores and early signs that customers are leaving.
- Win-back campaigns: find lapsed customers and send the segments straight to your email tool.
- An analytics warehouse on ClickHouse, with dashboards your team will actually open.
Proof: ClickHouse is the analytics database behind all my products, and I have years of e-commerce speed and SEO work behind me. Writing: posts about e-commerce →
Tell me which shop system and data you have. Email erik@datadepartment.io, call +31 6 52 49 89 64 or send a message.
How we can work together
- Free data consult. An hour-long video call to spar about your question, free of charge. Ask for one →
- Quick scan. A short, fixed-price look at your data or AI question, ending in written advice you can act on.
- Project. I build the system and hand it over with documentation.
- Data feed. I run the feed and keep it working each month, and you receive the data.
- Hourly. Extra hands for your team.
What I work with
Proven, open-source tools, chosen so that someone else can take over the system later.
- Programming languages: Python, Go and JavaScript.
- Collecting web data: Scrapy for crawling at scale, Playwright for websites that need a real browser.
- Storing and analysing data: PostgreSQL, ClickHouse for analytics, and Airflow to run the pipelines on schedule.
- Websites and apps: Django, FastAPI and WordPress.
- Servers: Linux and Docker, with Nomad or Kubernetes to run it all across machines. On your own hardware or a host you choose, and AWS where it is the right tool.