Niek - A young man with brown, messy hair, light facial hair, and a slight smile.
Niek - A young man with brown, messy hair, light facial hair, and a slight smile.

Niek Mereu

I build machine learning solutions that reaches your users.

Five years. Zero demos.
Everything I've shipped went into production.

STRV O2 PwC Miya STIL

What people call me for

Recommendation systems

Get the right content to the right users / recommend your users something they're

Elasticsearch · vector search · rerankers · online learning · PyTorch

Computer
vision

Deze text moet nog aangeleverd worden

PyTorch · fine-tuning · SageMaker · Lambda · ECS

GenAI
applications

make sure the LLM has relevant (company?) data within reach, so it can always answer your

RAG pipelines · vector databases · Bedrock · FastAPI

MLOps & productionization

make sure the LLM has relevant (company?) data within reach, so it can always answer your

RAG pipelines · vector databases · Bedrock · FastAPI

One man for the entire process

  1. 01

    Data

    Before any model, I go through what you already collect and what is actually usable.

  2. 02

    Model Data scientists stop here*

    I train on your data, not a benchmark set, and stop when it beats what you do now.

  3. 03

    Deployment

    The model goes somewhere it can run every day, with the infrastructure to keep it there.

  4. 04

    Integration

    Predictions arrive in a form your product can use, not a notebook someone has to open.

  5. 05

    Monitoring

    You hear it when the data changes shape, before your users do.

Most people cover one or two of these.

It’s why teams end up with three different specialists.

Niek Vs Them

Hiring in-house…

✓ Someone on your team full-time, long term

✗ Three months to hire, if you find the right one

✗ Salary, benefits and equity before a line of code

✗ You're betting one person covers all five stages

✗ A wrong hire takes a year to undo

Hiring an agency…

✓ A whole bench, and they can scale up

✗ Agency rates, with their margin on top

✗ You get whoever's free, not whoever fits

✗ Weeks of scoping before anyone writes code

✗ You brief an account manager, not the engineer who builds it.

Hiring me!

✓ Start in days, not months

✓ One rate, no margin on top

✓ Same person from data to monitoring

✓ Hire me for the whole project, or just for the part you're missing

✓ Clear documentation and a single contact point

See what working together looks like

Are we a fit?

Good Fit!

  • You have a product in production with real users

  • You know which problem you want solved

  • Your team can point at the stage they're stuck on

Not a Fit

  • You want the app built too…
    I integrate ML, I don't ship your frontend

  • You need someone to decide what your product should be

  • You have no data yet

  • You're using this as a trial run for a full-time hire

Don’t take my word for it

"

For STIL, Niek built a special ML ecosystem to identify and classify tremors. He lead the development of the full stack of a smartphone application, managed the ML, its implementation, production and testing pipelines, and hosting on our own infrastructure. Niek is an exceptionally capable, reliable, and pleasant ML engineer.

Nicola Pambakian

CTO, STIL

"

Working with Niek has been like cutting through butter with a hot knife… effortless, clean, and precise. He brings deep expertise in machine learning and data engineering, and consistently delivers clean and smart solutions whether that’s building scraping pipelines, wiring up Databricks workflows, or designing ML systems hungry for production data. What sets Niek apart is his ability to translate ambiguous business problems into robust technical implementations without over-engineering. He’s communicative, dependable, and a valuable partner on any data-heavy engagement.

Miya Consulting

Tell me what you’re building!

  • A reply within 48 hours

  • Projects run from two weeks to six months

  • Remote across Europe

  • Happy to sign an NDA

Book a Call