Details
Let me take you on a journey—a journey of bold ideas, spectacular failures, and the rare flash of brilliance.
It began in 2016 at a hackathon, where I naively believed I could build a shoe to help the visually impaired navigate the world.
Armed with an Arduino, an Android phone, and an unreasonable amount of optimism, I hacked together an obstruction-sensing prototype that actually worked… kind of.
That was my first real step into the open-source universe.
Since then, my open-source adventures have taken me through wild terrain.
From developing AI-driven NLP tools that untangle legal agreements (because who really reads those?) to training models that question media bias, I’ve seen firsthand how open-source collaboration fuels innovation.
Some projects were breakthroughs. Others? Glorious train wrecks. But each one taught me something.
In this talk, I’ll share real-world stories from these escapades, including:
- How unreadable Terms and Conditions emails led to a legal summarizer (PyCon India 2018)
- Trying to train a neural net to detect fake news—and questioning reality in the process (PyConZA 2019)
- The crucial role of bad demos in some of my best ideas (Only Bad Demos in the Building — LWT Summit & PyCon US 2024)
- Analyzing AI bias with SVMs (AI in Software Dev Summit 2024)
- Building a bullshit detector using LLMs (PyCon Estonia 2023)
- Chatting with ChatGPT about everything and nothing (PyCon Lithuania 2023)
- Investigating media polarization conditioning (PyCon Hong Kong, Sweden, EuroPython)
- Detecting subtle bias in children’s books (PyOhio 2020)