This website uses cookies
Read our Privacy policy and Terms of use for more information.
A two-week architecture audit for Fintech teams. You get one report: your five biggest risks, ranked, and the order I'd fix them in. No slideware, no retainer.
Tell me what's going onYou probably already have a hunch about where the trouble is.
There was an incident nobody wants a repeat of. The p99 has been climbing for two quarters. Someone on your team keeps flagging the same risk in planning, and it keeps sliding down the backlog because there's always something more urgent this sprint.
What you don't have is a straight, independent answer you can take to your board or your team and act on. I come in from outside, figure out what's most likely to hurt you and what it'll cost, and leave you with a plan you can start on.
Two weeks of work, and at the end of it, one document.
Inside: your five biggest risks, ranked. Not by how interesting they are to me, but by how much damage they'll do and how much it takes to fix them. For each one, I tell you what's happening, what I found, what it's costing you today, what it'll cost at three times the load, and what I'd do about it.
Then the part you're really paying for - the order to fix them in, and what has to happen before that migration you've been planning.
We finish with ninety minutes and your leadership team, going through the whole thing. You should push back. The report's better for it.
I don't write the code, and that's deliberate.
The minute I'm the one building the fix, I've got a stake in believing it's the right fix. You want the diagnosis from someone who doesn't. So: no pull requests, no pairing, no retainer, no standing meeting that eats your Thursdays, no small favour that quietly turns into a second job.
If what you need is more hands to ship features, I'm the wrong hire, and I'll say when we talk rather than take the work.
This works best if you're a CTO, a VP of Engineering, or whoever actually owns the platform, at a fintech or another regulated business with somewhere between 20 and 150 engineers. Your system's been around a few years and already survived one growth spurt. And you can get me read access to the code, a year of incident history, and three-quarters of an hour each with a few of your engineers.
It works badly if you've already decided what the answer is and you want someone credible to co-sign it. I'll give you my honest read, and it won't always be the one you were hoping for.
Eighteen years building software, eleven of them in fintech, on systems where being wrong about correctness or an audit trail isn't a mistake you get to make twice.
For most of that time, the thing people have called me in for is exactly this: walk into a system I didn't build, usually while something's on fire, and work out where it's about to give. Do enough of them, and you stop guessing. You just know the handful of places to look first.
I write up how I think about this every week over at Tech Architect Insights - retry storms, the data leak that never throws an error, the cloud bill nobody saw coming. Read two or three. If the way I reason about systems fits what you need, get in touch. If it doesn't, no harm done.
Tell me what's going onEvery audit is a fixed price, and you know it before I start. It doesn't creep up because the work turned out harder than I expected. That's mine to absorb, not yours. Half when we begin, half when you get the report.
The exact number depends on the size of your system and how quickly you need it, so we'll sort that when we talk. What I'll say now is that it costs less than a single week of the rewrite you're weighing up, and far less than the outage it's meant to head off.
Send me an email. A few lines is plenty - what's worrying you, roughly how big the system is, and whether something's already gone wrong.
I read them myself and reply within a day or two. If it sounds like an audit would help, we'll set up a half-hour call and talk it through properly. If I don't think it will help, I'll tell you that instead, and you'll have spent five minutes rather than an hour.
Useful to include, if you have it:
What's prompting this now - an incident, a migration you're weighing up, a new role
Rough size: how many engineers, how old the system is
The one thing you'd fix if you could stop everything for a month