Anyone interviewed at OpenZeppelin or Matter Labs lately? Let's decode their proof-based security tasks.

Just to kick things off—I was chatting with a hiring manager yesterday who mentioned that for smart contract auditing roles, a lot of protocols are heavily testing gas exhaustion attacks right now.

Has anyone had to write targeted unit tests to simulate DoS via gas limits during a live technical interview, or is that mostly just in the take-home portions?

I went through the OpenZeppelin smart contract auditor process recently, so I can share exactly what their paid work trial looks like right now.

To answer your question directly: no, they are definitely not handing out Ethernaut CTF puzzles anymore. The trial is designed to emulate an actual client engagement.

Here is what caught me off guard and what they actually test:

1. The Codebase: They hand you a near-production, highly realistic codebase (think a fork of a DeFi protocol with custom logic layered on). You get a few days to review it on your own schedule, and yes, they actually pay you for your time regardless of the outcome.

2. Tools vs. Manual Review: You can use Slither or any static analysis tool you want, but they will only catch the low-hanging fruit. The critical/high-severity vulnerabilities they seed in the code are almost always tied to the protocol's specific business logic. You have to manually spot things like a subtle mathematical precision loss over time, or a read-only reentrancy that manipulates a specific state invariant. A tool won't find that for you.

3. The Real Grading Metric: Finding the vulnerability is only half the battle. They grade you heavily on your reporting and mitigations. If you find a critical access control bypass but your suggested fix breaks the core functionality of the protocol, you fail. You need to write a clean, professional report and back it up with a working Foundry Proof of Concept (PoC).

If you treat their trial like a bug bounty and just submit a disorganized list of 15 minor gas optimizations, you won't pass. They are testing if you can act as a professional security advisor to a client, not just a hacker.

Curious if anyone else here has taken it recently—did you get a DeFi protocol, or was your trial codebase more focused on governance/infrastructure?

Here’s a look at how my interview with Matter Labs went. If you're targeting the ZK Engineer role, skim-reading the PLONK paper isn't going to cut it.

The technical screen is tough, especially if you're used to relying on high-level frameworks without really understanding the underlying math. They drill deep into circuit security. What caught me off guard was their focus on arithmetic underflows over a prime field. In short, if you're writing in Circom and miss a constraint, a malicious prover can forge a perfectly valid proof for an invalid state. They actively test whether you can catch under-constrained circuits in real-world scenarios.

Beyond that, you need to show you can optimize for the prover. It’s all about balancing ironclad security with keeping proof generation computationally feasible by keeping your constraint count low.

My biggest takeaway for anyone prepping: don’t just practice writing circuits—practice breaking them.