velodata.ai velodata.ai The Mathematical Beauty of Cycling
  • Blog
  • Projects
  • Apps
  • About
  • Work with me

Data science, applied maths and modelling for professional cycling.

I help cycling teams, brands, race organisers and sports-tech companies answer their questions with data: how hard a stage is for each rider, whether a product claim holds, why riders abandon a race.

Work with me Read the blog

Portrait of Juanfran Garamendi

I’m Juanfran Garamendi. I hold a PhD in Computer Science & Mathematical Modelling and have spent 15+ years working in data science and applied mathematics, in academia and industry. I’m also an ultracyclist, and I founded velodata to bring the two together.

More about me

Who I work with

Teams

How hard each stage is for each rider, who to line up, and which equipment is really faster.

Brands

Whether a product claim holds, and how to design the test that proves it.

Race and ultra-race organisers

Reports about your race: heat and weather, abandonments and course difficulty.

Sports-tech companies

Data science, machine learning and mathematical modelling projects made to measure.

See what I offer

Latest posts

Strip plot of the standard deviation of the watts per speed ratio across my repeated rides. Most points sit below 0.5 and a few reach 1.4.

Tubeless vs. Tubular: Are SRAM’s Claims Inflated?

equipment

Let’s delve into SRAM’s experiment with the Movistar cycling team. This article explores whether the methodology genuinely demonstrates the superiority of tubeless tires or if there is more to the story.

Oct 1, 2026
No matching items

All posts

Projects

Heat across five editions of an ultracycling race
Finished

Which edition of the Madrid-Barcelona by Pedalma was the hottest, and where on the route does the heat hit? A comparison built from my own GPS files.

DNF risk in self-routed ultracycling
In progress

Do women and men abandon differently in the Transcontinental Race? A survival analysis of 568 solo riders.

A physics-based stage difficulty score
In progress

A stage difficulty score based on physics and specific to each rider, meant to replace gradient-only profile scores.

No matching items

All projects