About
I'm a Senior Product Manager based in Cape Town, currently at TurnStay — a VC-backed seed-stage startup building multi-currency, cross-border payment infrastructure for the travel industry. I joined as the founding PM, working directly with the founders on product strategy, go-to-market, and standing up the internal tooling to make the company run.
Before TurnStay, I spent nearly five years at Prodigy Finance, a Series C fintech providing international student loans. I started as a Junior PM and grew into leading two domains: Customer Platform (authentication, comms, support infrastructure) and Verify & Disburse (KYC, AML, loan fulfilment). The work ranged from migrating core authentication systems to shipping AI-powered tools for document verification and complaint detection — a few of which you can read about below.
My background sits somewhere between product, data and engineering. I studied Economics at UCT, taught myself Python and SQL along the way, and have always been drawn to the messy, interesting problems at the intersection of business logic and technical systems — especially where automation and AI can do the heavy lifting.
Outside of work: avid gamer, tech enthusiast, and always happiest somewhere outdoors.
Portfolio
At Prodigy, we treated customer support as a competitive differentiator — but a rising trend in complaints signalled a deeper issue. Our analysis revealed that most complaints weren't isolated incidents; they were the result of accumulated negative interactions over time. Three consistent markers stood out: customers having to repeat themselves, the tone and sentiment of their messages, and the urgency they conveyed.
Rather than solving each potential root cause, we focused on detecting when a customer was veering into a negative interaction path early enough to intervene before frustration escalated into a formal complaint.
To address this, we built Sentinel — an AI-powered early warning system. Using Zendesk webhooks, OpenAI's API and a custom internal service, we analysed ticket history and context in near real-time to classify sentiment, urgency and repetition. These insights were surfaced directly in Zendesk via a custom app, empowering agents to prioritise and escalate at-risk conversations. We rolled out Sentinel incrementally, refining our prompts with a test group before expanding to full production — scaling to thousands of tickets daily. This not only improved customer experience but gave CS leadership early signals of broader product or process issues.
Manual document review was a significant bottleneck in Prodigy's KYC process — labour-intensive, error-prone and difficult to scale. Working closely with the Data Science team, I led the product development of a multi-agent AI verifier capable of reviewing uploaded documents, cross-referencing data and flagging exceptions for human review.
The result was a 50% cut in manual verification workload, faster customer onboarding, and a more consistent, auditable review process.
Prodigy's UK market was underperforming relative to its potential. Research revealed that the existing disbursement flow was a barrier to higher-value loans — funds were routed through university finance offices rather than directly to students, creating friction and limiting the addressable loan amount.
I led the design and launch of a direct-to-customer disbursal feature that bypassed this constraint. The result was a 61% increase in average UK loan size and a meaningful gain in market share in non-US markets.
Skills & Tools
