I am Santosh Srinivasaiah. I work as a Senior Lead Scientist on precision medicine, brain-computer interfaces and semantic search, which in practice means I spend most of my time between two groups of people who rarely read each other’s papers: the ones who study brains and the ones who build software. This site is the translation guide I wish had existed when I started.
Why this site
Computation has become essential across every science, for acquiring data, analysing it, automating experiments, and testing hypotheses through modelling and simulation. Software development is now a scientific activity. Yet training scientists to program, choosing the right tools and making them work together are still treated as afterthoughts, and science is slower and less reproducible for it.
Neuroscience feels this more sharply than most fields. Recordings have grown from a handful of neurons to tens of thousands, and the people generating the data often cannot analyse it themselves. At the same time, machine learning has drifted away from the brain that inspired it, and most of its practitioners have never looked at a spike train. This site tries to close both gaps at once: a learning path that teaches programming through real neuroscience, from the first line of Python to the point where you can build a decoder, a model of a neuron, or a brain-computer interface yourself.
What you will find here
- A structured path of seven tracks, from Python basics to decoding and clinical machine learning. The front page has the map; the Foundations track is complete and the rest are being written in order.
- Lessons built on real, openly licensed neural data, with the complete code and its actual output. Every number and figure in a lesson comes from running the code you see; nothing is illustrative.
- Longer essays on where neuroscience and artificial intelligence meet, in the Neuro meets AI category.
- A curated, link-checked list of open data sources for neuroscience and health.
How the lessons are made
Each lesson is a single script that runs top to bottom. I write the analysis first, make sure the result is real and the figures are honest, and only then write the words around it. Where a result has a caveat, the caveat is in the lesson, because the caveats are where most of the learning is. The data comes from the sources on the data page, mostly the Steinmetz recordings shared by Neuromatch Academy, and is always credited.
You do not need a neuroscience degree or any programming experience. You need Python, or a free notebook in the browser, and about an hour a week.
Using this material
The text, code and figures on this site are free to use for teaching and learning; a link back is appreciated. The datasets have their own licences, noted on each lesson and on the data page, and must be cited separately. If you build a course or a workshop on these lessons, I would be glad to hear about it.
Get in touch
Corrections, questions and suggestions are welcome, and corrections especially. Find me on Twitter at @thinquanaut or through the contact page. My ORCID iD is 0000-0002-5444-047X.
The views here are my own and not those of my employer. I have no conflicts of interest to declare in relation to anything published on this site.