Benedikt Bünz ☕️ (@benediktbuenz) 's Twitter Profile
Benedikt Bünz ☕️

@benediktbuenz

Chief Scientist @espressosys ☕️ (WE ARE HIRING)
Assistant Professor Computer Science @NYU_Courant

ID: 795521517433929728

linkhttp://bue.nz calendar_today07-11-2016 07:01:24

2,2K Tweet

10,10K Followers

765 Following

Benedikt Bünz ☕️ (@benediktbuenz) 's Twitter Profile Photo

Me: MRI in the US with insurance: $6,464.38, cost to insured $1821.35. My brother in Belgium. Full blood work: 32€ , cost to insured: 3.20€.

Benedikt Bünz ☕️ (@benediktbuenz) 's Twitter Profile Photo

The bears are back in town -> time to focus on buidling. Let's make sure to build something real this time and focus on end-user needs. Solid infrastructure that is transparent to end users vs. leading to infinite drop down menus, compelling applications, good ux! We can do it.

Benedikt Bünz ☕️ (@benediktbuenz) 's Twitter Profile Photo

Hardware acceleration and Hardware <> Algorithm codesign can deliver 1000x speed-ups and usher in the next era of zk proving. Had a great time helping Alhad Daftardar , Jianqiao, Joey, Ramesh, Siddharth Garg 🌈, and Brandon accelerate HyperPlonk. To appear at ISCA: eprint.iacr.org/2025/620.pdf

Benedikt Bünz ☕️ (@benediktbuenz) 's Twitter Profile Photo

Only devnet but things to look forward to ☕️. Main improvement is the lazy VID we described here (original idea usenix.org/conference/nsd…): espresso.discourse.group/t/zoda-frida-d…

Daniel Escudero (@deescuderoo) 's Twitter Profile Photo

Latincrypt 25: FINAL DEADLINE EXTENSION You now have until Fri May 16th (AoE) to submit your papers to Latincrypt 2025. Latincrypt takes place on Oct 1-3 in Medellín, Colombia. Airports are well connected, and the capital Bogotá (45mins flight) even more!

Latincrypt 25: FINAL DEADLINE EXTENSION

You now have until Fri May 16th (AoE) to submit your papers to Latincrypt 2025. 

Latincrypt takes place on Oct 1-3 in Medellín, Colombia. Airports are well connected, and the capital Bogotá (45mins flight) even more!
Benedikt Bünz ☕️ (@benediktbuenz) 's Twitter Profile Photo

Super excited to see all the progress in real time proving! It’s highly compatible with espresso’s confirmation layer. We provide consensus on the transactions. Today we are using TEEs to get to from consensus of transactions to consensus of state. In the future we can use proofs

Ben Fisch ☕️ (@benafisch) 's Twitter Profile Photo

Succinct summary: while too expensive on Ethereum L1, today CLOBs can run as L2s using a cheaper base layer for availability and ordering of data (e.g. Espresso or Celestia) and still settle to Ethereum. Plus as a ZK validium orders on the base layer can be encrypted for privacy.

Benedikt Bünz ☕️ (@benediktbuenz) 's Twitter Profile Photo

Super excited to teach about folding and accumulation at the ASCrypto Fall school. If you want to learn about the latest and greatest in proof systems come join us in Colombia 🇨🇴

Kartik Nayak (@kartik1507) 's Twitter Profile Photo

Pranav Garimidi This is a lovely article! Here's a recent (small) improvement on one of the aspects: eprint.iacr.org/2025/279.pdf (with Dan Boneh, Lior, Victor, Benedikt Bünz ☕️). We demonstrate that we need not incur latency penalties when decrypting using committees; one can decrypt even before

Alireza Shirzad (@alrshirzad) 's Twitter Profile Photo

Check out our new paper: IronDict — Transparent Dictionaries from Polynomial Commitments 🔑 1B entries audited in 35ms with an 8KB proof 📉 300× faster auditer, 150,000× smaller proofs than WhatsApp's AKD 📄 eprint.iacr.org/2025/1580 💻 github.com/alireza-shirza…

Check out our new paper: IronDict — Transparent Dictionaries from Polynomial Commitments

🔑 1B entries audited in 35ms with an 8KB proof
📉 300× faster auditer, 150,000× smaller proofs than <a href="/WhatsApp/">WhatsApp</a>'s AKD

📄 eprint.iacr.org/2025/1580
💻 github.com/alireza-shirza…
Benedikt Bünz ☕️ (@benediktbuenz) 's Twitter Profile Photo

Hydranga is a super cool consensus protocol where you can distinguish between fully Byzantine and crashed/non-responsive notes. This enables a fast path when (almost) everyone is online

Benedikt Bünz ☕️ (@benediktbuenz) 's Twitter Profile Photo

Fed the same small data analysis task to ChatGPT, Gemini and Claude and got 3 different answers. Then did it myself with excel and got a 4th (and hopefully correct answer)