Full-stack & machine learning engineer

Jeronimo Muñoz Larreta.

I build software and work with artificial intelligence.
I design websites for businesses.
Choose what you’d like to explore.

Projects and technical experience

Software &
artificial intelligence.

View experience & CV

Web design and development

Websites.
With an identity.

View websites
Two sides. One way of building.Or keep exploring

Applied machine learning

From a signal.
To an output.

I work with radio signals: preparing data, extracting features and comparing models. Explore selected public samples and follow how their archived outputs were obtained.

DroneRF · Radio-frequency signalsA walkthrough of archived data
  1. Input signalRF fragment16,384 signal samples
  2. Signal measurements289 featuresTemporal and spectral
  3. Saved output—HGB · MLP · Random Forest
The fragment’s signalNormalized amplitude
512 visible samples of the original signal, normalized.

Preparing the fragment…

View the time-frequency representation
Representation of a public DroneRF signal.

Frequency ↑ · Time →
Relative energy and normalized axes.

Model votes

The output is distributed across four classes.

Archived mean votes. These are not calibrated probabilities.

Loading archived results…

Selected public samples with archived outputs.
The animation explains the process; the browser does not run the classifier.

Explore the full case
Data and method

DroneRF, Mohammad Al-Sa’d, Mhd Saria Allahham, Amr Mohamed, Abdulla Al-Ali, Tamer Khattab and Aiman Erbad (2019), DOI 10.17632/f4c2b4n755.1, CC BY 4.0. Visual transformation and archived evaluation by Jeronimo Muñoz Larreta. These selected samples are not an overall metric or an operational field test. Outputs come from an HGB + MLP + Random Forest refit, not historical CNN weights. The classifier input has 16,384 samples; the waveform shows the first 512 normalized display samples. The time-frequency representation contains 129 × 64 cells with energy normalized per fragment; values are not calibrated dBm. The full case and its evaluation are available in my CV.

Your next project can start here

What could we
build together?

Software, artificial intelligence, websites
and work opportunities. Tell me what you have in mind.