talks

Poster session. By using Python libraries like Librosa and PyTorch, we analyze audio at a level beyond human perception, detecting subtle synthesis artifacts that reveal synthetic origins.

Poster session. How machine learning is used to predict the properties of materials before anyone makes them: representing crystals and molecules as data, the models that work on them, and what Python’s scientific stack makes possible for a lab without a supercomputer.

My first TEDx, for the theme “The Space Between”. The argument is simple: the useful part of life is the bit between knowing and not knowing. I moved to Germany with two words of German. I started Gezellig by inviting a few friends over on my birthday. Neither was a plan. Action comes before motivation, and the small, scary step you have been avoiding is usually the one that matters.

Pygame fundamentals, then PygBag: compiling a Python game to WebAssembly so it runs in any browser with no install. Live-coded from an empty file to a small playable game on the stage.

A three-hour hands-on tutorial. Threads, processes and concurrent.futures first, then asyncio from the event loop up: what I/O-bound and CPU-bound really mean, when each tool is the right one, and the mistakes that make async code slower than the synchronous version. Prerequisite: basic Python.

Charts are not stories. This talk combines Blender’s 3D animation, scripted from Python, with the 2D plots you already make in matplotlib, Plotly and seaborn, to turn a dataset into a narrative that moves.

Python is friendly until it isn’t. A live tour of the language’s surprising corners: mutable default arguments, late-binding closures, integer caching, is versus ==, floating-point traps, and the odd behaviours of datatypes and syntax that bite even experienced developers. Every example runs in front of the room, the reason is revealed afterwards. I gave this one in six countries over three years.

A hands-on 3-hour workshop for people who know some Python and no data science. NumPy and pandas from a blank notebook, loading a real dataset, cleaning it, asking it questions, and plotting the answers.

Everything I wish someone had told me before the first meetup. Venues, speakers, timelines, the month nobody shows up, the communities that grew out of it, and how a conference grows out of all that. Ends with how to start something similar in your own city, and the mistakes to skip.

A lightning introduction to NumFOCUS, the PyData meetup network, and how a city can start its own chapter.

The basics of gathering data from the web: pages and APIs, submitting forms, dealing with JavaScript-driven DOMs, and a side-by-side of BeautifulSoup, lxml and regular expressions. It ends with the ethics: when you should, when you shouldn’t, and how to scrape without getting banned. The first version of this was my first conference talk ever, and my first international one as well.

I chaired the first PyData conference in India, at IIIT Delhi, eleven months after founding the meetup. Two days, two tracks, and a team of five organisers, and about 500 people.

How PyData Delhi went from an idea to thousands of members with no budget and no venue, and what that taught me about why people show up, why they stay, and how newcomers learn fastest from the people around them. My first keynote, at twenty, sharing the slot with Jeff Knupp, which I did not find intimidating at all.

The first PyData Delhi talk. A short talk on what data science actually is and why New Delhi needed its own group for it. The group passed 1000 people within six months.