Back to our Developer Showcase Series to learn what developers in the real world are doing with LF Decentralized Trust (LFDT) technologies. Next up is Roman Vaseev, Staff Software Engineer at Consensys.
I work on several components of the Lineth stack, the zk-rollup stack that powers Linea: the Coordinator, which sequences transactions and orchestrates proving; Maru, the consensus client; the rollup guest programs; and some of the Lineth Besu plugins. A zk-rollup executes transactions off Ethereum and posts one cryptographic proof back to it, so Ethereum verifies a proof instead of re-executing every transaction. That is where the cost reduction comes from.
I got into decentralized technology through Consensys in 2021. We were building a proof of concept around zk-rollups. It became Linea, and most of it is now part of the LFDT project Lineth. I stayed for the combination of factors: high complexity, cutting-edge research, other strong teams building similar things in the open, and Ethereum values that happen to be mine. The direction and emphasis have shifted several times since 2021. It is still a demanding project, and still a fun one.
Most of my contributions go to the Lineth repo, with occasional PRs to Besu and Web3j, since we depend heavily on both. I spend more time in pull requests than in chat channels.
Lineth was being built in open source before it moved to LFDT, so contributing was not a new decision. It was already how we worked. What LFDT changes is who else is in the room. Besu is there, and so are the institutions running this software in production. When a zk-rollup stack sits in the same foundation as the client it is built on (Besu), some decisions get made once instead of twice.
Besu is not a fair example as Consensys has a team dedicated to its development too. Web3j is the better one. The pattern is usually: an Ethereum fork adds a feature, we need it, Web3j does not have it yet, we open an issue or send the PR ourselves, it gets reviewed and released. Some steps take days. What strikes me is that none of it requires an agreement between companies. In a closed setup, we would be negotiating. Here we are reviewing code.
I look at this from the perspective of Ethereum values. Building cryptography in the open advances the field faster than any single team can. When the good architectural decisions are visible, you can combine them for your use case instead of rebuilding everything and compromising along the way. The same applies to research: our cryptography team publishes its work, and gnark, the zk-SNARK library behind our prover, is open source for the same reason. A proof system nobody can inspect is not worth much.
Besu. Forking Geth or Reth is always an option, but Besu gives us cohesion with the rest of our stack, real code reuse, and robust and reliable testing harness. The Lineth Besu plugins exist because of that: the L2 stack extends the client instead of working around it.