RockWallet
This one did not start with an empty repository. It started with a half-working clone of somebody else's wallet, with Android, C and Swift sources tangled together in the same project. Turning that into a regulated product meant taking it apart before adding anything to it — and then carrying it through two brands.
What we inherited
The starting point was a clone of a repository derived from Breadwallet / BRD, the open-source crypto wallet. It was half functioning, and it was a mess. Android sources, C sources and Swift sources all lived together in the same repository, without clean separation between the platforms or between the wallet core and the application built on top of it.
That is a specific and unglamorous kind of problem. Nothing is missing exactly — there is code, and some of it works — but you cannot reason about any one part in isolation, so every change is a risk and no one can estimate anything honestly.
Taking it apart
- Went deep into the inherited codebase to understand what was actually there, rather than working around it.
- Decoupled the tangled sources — separating the platform code and the wallet core into something that could be worked on independently.
- Established a codebase the team could then build features into with some confidence.
This was the precondition for everything else. The feature work that followed was only possible because the foundation stopped fighting back.
Two brands, one codebase
The product shipped first as Fabriik. It was later rebranded to RockWallet — the same lineage carried through a change of identity rather than restarted. Beyond the rebrand, the application grew well past what the original open-source wallet did:
- Wider crypto asset support than the inherited wallet handled.
- Trading, and buying and selling flows.
What the product is
RockWallet is a regulated crypto wallet — buying, holding and swapping assets, with KYC/AML-compliant account setup, identity verification and biometric authentication. Operating in a regulated context sets the constraints that everything above had to satisfy.
Why this one is worth reading
"Built a crypto wallet" would be the easy summary and it would be the wrong one. The work was inheriting a codebase nobody could safely change, making it tractable, and then taking it through a rebrand and a significant feature expansion without losing the thread. That is a more common situation in real engineering than a green field, and a harder one to do well.
Stated precisely: the dates above are approximate. KYC/AML onboarding, identity verification and biometric authentication are described here as properties of the product, not as work claimed personally — how much of the regulated-compliance implementation was mine is not something I want to overstate from memory.