Wargame strategies for handling user misbehaviour in mass B2C platforms.

If you run an app where millions of users can contribute ANY form of content, you are in the daily business of keeping thousands of crazy internet people from destroying your brand and / or exposing you to extinction-level legal risk.

I've handled these issues at high scale for international companies with hundreds of millions of users, and I now have an approach that works both for executive buy-in and pragmatic solutions in the field.

In this session I walk through two case studies and offer a practical approach to get your executive team involved and prepare for the worst that can happen, because it will.

Content note: some spicy topics will be discussed, but there will be no spicy visuals and no graphic descriptions.

Business and customer insights from building large scale recommender systems

Recommender systems are strange and fascinating. I’ve built and consulted on recommender systems for the some of the largest retailers in South Africa, and this talk is about all the weird stuff that goes on while you’re just trying to help people buy groceries.

Your customers will be offended by recommendations that are mathematically true, odd products will dominate the feed for even odder reasons, and your code will suddenly become the focal point of some serious ethical issues.

In this session I share all those stories, including some smart ways to learn about your customer, and how use that information in safe and ethical ways.

Happy, productive, remote: international technical teams and how to manage them

Managing technical teams spread across different countries is a challenge worth conquering. This is an operating system for bringing people together around shared, complicated tasks, when all you have is Slack, git and video meetings. The work itself is already hard, good team people are hard to find - but how you manage the team determines how much everyone can bring to the table.
I've learned some stuff the hard way, and developed my own techniques to handle teams - I'm sharing my operating system.

When a Recommender Fails

What happens when the algorithm doesn’t deliver? In this honest, behind-the-scenes session, we unpack the technical,
organisational & human challenges that led to a failed recommender system project. From messy data to misaligned goals, learn
what went wrong & what we’d do differently next time.

First presented at DataFest Cape Town 2025
About 20% technical talk about the underlying computational issues, majority of the content is about human issues, biases, organisational challenges, what we learned and how to lead through something like this.
20-30 minutes, simple slides, no audio or video needed.
Target audience is data teams and leaders of data teams or departments.