We present results relevant to front-running protection systems based on threshold cryptography, which are currently eliciting interest. They are being implemented and deployed actively and will likely be part of Endgame Ethereum and rollups.

To achieve 10x higher throughput in Ethereum, we need a sharded blob mempool and data availability sampling. We explain a design for fast and efficient blob dissemination and data sampling in the p2p networking layer.

We model the expected proposer-revenue effects of various preconfirmation protocols, as well as discuss the knock-on effect of preconfirmation revenue on the proposer set. We also propose a pricing model for preconfirmations.

EVM.lua is a research project that embeds an Ethereum Virtual Machine interpreter within Redis using Lua-based Functions. By enabling Redis to directly interpret EVM opcodes, this innovation aims to drastically reduce processing time of eth_calls.

ETH has a marketing problem. Technology is adopted in S curves and the next cohort are $trillions of institutional capital. Crafting the right bridges to onboard TradFi capital will unlock price action and mindshare, resulting in net new users and catalyze the next step function of innovation on Ethereum.

Presenting the main ideas, results, and key technical details from our work, 'Forking the RANDAO: Manipulating Ethereum’s Distributed Randomn' co-authored by István Seres, János Tapolcai and Bence Ladóczki: https://eprint.iacr.org/2025/037.pdf.

Home staking is the ideal but often unattainable Ethereum staking model. This talk explores the technical and economic barriers to home staking and how how innovations like CSM can help to build a more resliant and decentralised network

Vero is a new validator client that can aggregate data from multiple beacon nodes running different client implementations. This significantly reduces the impact of any single client bug, making Ethereum more resilient while also reducing risk.

In this talk, I will present cryptographic tools for proving the correctness of inferences made by popular deep learning models. These tools leverage advancements in SNARKs to achieve significantly faster prover and verifier times—up to two orders of magnitude faster than an existing baseline.

Learn how Zircuit Prover's modular design accelerates ZK integration, reduces costs, supports diverse proof systems, and optimizes performance

Binary trees are making a strong return as the preferred solution for implementing stateless Ethereum. This talk will go over the history, design trade-offs and future potential of replacing verkle with binary trees.

Most of MEV's profit has been distributed to centralized entities in many EVM blockchains. We propose a new auction-based MEV protocol and its implementation for high-performance EVM blockchains.

I will first discuss when zero-knowledge is useful from computational efficiency perspective, and then briefly go over how modern zero knowledge systems work. I will then use some examples like using randomness from Fiat-Shamir to optimize circuits to illustrate ​some optimization techniques, and if time allows, discuss results and unlocked applications with better performance.