Welcome back to I'm a Software Engineer ~ What Next? The podcast for devs figuring out their next move. Hosted by James Wilson and Matt Sinclair, we talk honestly about what it takes to build a meaningful career in tech.
This week we're joined by Sean Hederman, CTO of Zilch, one of the UK's biggest fintechs. Sean was the second hire at Zilch. Before that he built bi-temporal reference data systems at the Johannesburg Stock Exchange, transformed DevOps at Stanlib and Direct Line, and somewhere along the way worked out that being the best individual engineer in the room is a ceiling, not a career.
In this episode, we cover:
- Why "being right doesn't scale" and what to engineer instead
- The multiplier effect: lifting a team by 20% beats doubling your own output
- Conway's law and the reverse Conway manoeuvre at Zilch
- Queuing theory applied to engineering teams (and why Sean mandates 20 to 30% tech debt work)
- Humans as chaos monkeys, and why half of software engineering practice exists because we're unreliable
- Hiring engineers in the agentic era and the AI usage patterns Sean actually looks for
- Adversarial agentic coding, spec-driven development, and getting the model to review its own work
- The myth of the 10x programmer and what real force multipliers look like on a team
⏱️ Chapters
- 0:00 Intro: Vietnamese Frogger and the move back to the UK
- 0:46 Meet Sean Hederman: Visual Basic 3, then the JSE's bi-temporal reference data
- 4:37 Stanlib: from 2 to 144 releases a year, then DevOps and cloud at Direct Line
- 6:23 Employee two at Zilch: a mission for the underserved and founders who execute
- 10:21 Ideas are like backsides: no helicopter money, and the counter-offer mistake
- 13:36 Seven suitcases to London, then the magic telephone: being right doesn't scale
- 16:52 Where do people fit in? Humans stay meaningfully in the loop
- 18:30 The multiplier effect: lifting a team by 20% beats doubling your own output
- 21:39 Engineer or leader? Two kinds of CTO, and standards that save conversations
- 24:38 Conway's law and the reverse Conway manoeuvre: teams that should not talk
- 28:05 Queuing theory: why Sean mandates 20 to 30% tech debt work
- 29:32 Humans as chaos monkeys, and Distra: reliable software on unreliable hardware
- 33:29 Hiring in the agentic era: passion, and the AI usage patterns Sean looks for
- 35:47 AI for the APIs you never wrote, declarative Ash, and scoping non-determinism
- 39:21 Opus 4.6, adversarial agentic coding, and spec-driven development with Kiro
- 42:54 Dividing the room on functional programming, Haskell, and the two TDD extremes
- 47:39 New engineers: excited or worried? Both, and Jevons paradox means more software
- 50:08 Is AI the Industrial Revolution? A soft landing, and AGI unevenly distributed
- 53:16 Never say never: agents have no agency, and high dimensionality beats intuition
- 55:11 The myth of the 10x programmer: 2.5x output plus 20% on everyone around them
- 57:27 The real force multiplier: lowest commit count, biggest effect on the team
- 59:23 Wrap-up: the Zilch journey, next time
🔗 Links
🗣️ Guest
- Sean Hederman - https://www.linkedin.com/in/sean-hederman/
🎙 Hosts
🌐 More from us
🎬