Detection & Counting
Finding and counting objects in crowded, messy images — microscopy cells, scientific captures, anything where manual counting breaks down.
Available for freelance projects
Etele Kovács — Computer Vision & ML
I build computer-vision and ML systems that take raw data all the way to working predictions.
Freelance computer vision and applied ML. Right now that means counting pico-algae cells in microscopy images and turning raw underwater light readings into water-quality datasets — with the code, data, and results all in the open.
What I Build
What I can build for you.
Detection, classification, and full data-to-deployment pipelines — shaped around your data and constraints, not a generic template.
Classification
Turning structured or visual data into reliable predictions, with evaluation that holds up outside the training set.
Segmentation & Analysis
Image-analysis pipelines built to be inspected: clear outputs, reproducible runs, and context that fits your domain.
End-to-End ML Systems
The whole path — preprocessing, feature engineering, APIs, metrics, and a project structure that's ready to deploy and hand off.
Process
What working together looks like.
A short, predictable path from problem to something you can run — with honest checkpoints along the way.
- Step 01
Scope & data check
We start with the problem and your actual data. If machine learning isn't the right tool for it, I'll tell you up front rather than build something that won't hold up.
- Step 02
Build & evaluate
I build the pipeline and model, measure against metrics that match how the result will be used, and share progress as it takes shape — no black box at the end.
- Step 03
Deliver & hand off
You get documented, reproducible code and a way to run it: an API, an interface, or a clean repo your team can pick up and extend.
Featured Projects
Two projects, documented end to end.
A 6-channel detector that counts cells in microscopy images, and a scientific pipeline that turns raw field readings into model-ready data. Real repos, real metrics.
Risk & Tabular ML
Automobile Insurance Claims Risk Modeling
Predicting motor-insurance claim frequency on 678K French policies — with a decision tree, neural network, and PCA implemented from scratch and benchmarked against a Negative Binomial GLM.
Result Snapshot
Policies Analyzed
678,013
Group ProjectMachine LearningFrom ScratchTabular Data
Case studyOpen
Microscopy Image Analysis
DAPI Microscopy Biomass Pipeline
A proposal-to-biomass pipeline for DAPI fluorescence microscopy: segment at high recall, review and refine, then measure biovolume from clean contours.
Result Snapshot
Status
In progress
In ProgressComputer VisionImage SegmentationMicroscopy
Case studyOpen
Microscopy Image Analysis
Pico-Algae Detection and Counting
A microscopy detection pipeline that counts pico-algae cells from paired brightfield and fluorescence image channels.
Result Snapshot
Processed Samples
250
Applied MLComputer VisionObject DetectionMicroscopy
Case studyOpen
Environmental Data Science
Spectral Attenuation Pipeline
A scientific data workflow for deriving optical attenuation outputs and related light-penetration indicators from field measurements.
Result Snapshot
Test R²
0.86
Environmental Data ScienceScientific PythonWater Quality ModelingSpectral Analysis
Case studyOpen
Work Explorer
Compare the projects at a glance.
Switch between projects to see the problem, the key numbers, and a link straight into the full case study.
Risk & Tabular ML
Automobile Insurance Claims Risk Modeling
A machine-learning study on the French motor third-party liability dataset (~678,000 policies) that predicts how many claims a policy will file in a year. A decision tree, a feed-forward neural network, and PCA were implemented from scratch in NumPy, validated against scikit-learn and PyTorch references, and benchmarked against a Negative Binomial GLM — the actuarially natural model for over-dispersed count data.
Policies Analyzed
678,013
French motor third-party liability dataset.
Models Compared
6
From-scratch DT & MLP, sklearn DT, PyTorch MLP, Random Forest, NB-GLM.
Group ProjectMachine LearningFrom ScratchTabular DataRisk ModelingRegression
Open case studySkills & Stack
The stack behind the work.
Every tool here is one I've used to ship the projects on this site — not a wishlist.
Vision & Modeling
PyTorchTorchvisionOpenCVscikit-learnCatBoost
Data & Scientific Workflows
PythonPandasNumPySciPyMatplotlibpvlib
Interfaces & Delivery
FastAPIPydanticReactViteTypeScriptVercel
About
How I think about the work.
Serious engineering, documented plainly — so you can see exactly what was built and how it holds up.
I take computer-vision and applied-ML problems from raw data to something that runs — and I document exactly how each project works, what it measures, and where the tradeoffs are.
My approach
Reproducible pipelines and clean interfaces that make the work easy to inspect, trust, and build on.
Read moreContact
Like what you see? Let's talk.
Have a computer-vision or ML problem you need solved?
Tell me what you're working on — detection, classification, a data pipeline, or something earlier than that — and I'll tell you honestly whether I can help and how I'd approach it.
Get in touch