Saravanan Vijayakumaran

@gm_usi I agree that the proving time is too high for acceptable UX. To improve, I need to switch to a scheme that supports lookups. Dorian, Nebula, NeutronNova are possible candidates.

@sarva_v 12MB is great for mobile use cases 👏 50s proof generation, though, will let users churn. Their avg. attention span is in the low seconds range. Hope you can find further improvements

Unpopular opinion: client-side ZK is nowhere near ready. Tried a popular ZK app today: Download a 400MB wallet. Oof. Launch app: hit with another 300MB download for prover artifacts. Wifi’s melting. Create first tx. Phone goes into overdrive generating the proof. 15+ secs of pure grind before I can even hit “submit”. The UX is brutal. I’m already considering churning… and I haven’t even started using the app.

They’ve authored protocols, built cryptographic systems, and shaped the research agenda. This summer, they teach at IIT Bombay. ACM India Summer School on Cryptography 📍 June 2–12 | Hosted by Trust Lab Apply now. #IITBombay #Cryptography #PQCrypto #Cybersecurity #ACMIndia

Excited to share that my master’s thesis work — "MProve-Nova: A Privacy-Preserving Proof of Reserves Protocol for Monero" — has been accepted at PETS 2025. This is joint work with my thesis advisor, Prof. Saravanan Vijayakumaran (@sarva_v ).

Recently summarized how NeutronNova improves on prior work—thought it’d be useful to share it. Whether you're working on recursive proofs, zkVMs, or just want a modern folding scheme—NeutronNova is worth watching. It achieves the best of prior work—without their downsides.🧵👇

@zkproofs @EF_ESP I emailed them at academic-grants@ethereum.org. They said they will reactivate the submission link soon.

🪄 Big ideas need time, so we're giving you more We've extended the deadline for the 2025 Academic Grants Round to March 23, 23:59 AoE! Take your research to the next level and apply now 🌟 esp.ethereum.foundation/academic-grants

Just dropped a blog post breaking down the circle group and trace domain for Circle STARKs! Stay tuned for Part III, where we’ll dive into the circle FFT.

Introducing Neo—not the One from the Matrix, but a new folding-based proof system, with: ✅ Efficient recursion (like Nova) ✅ Small fields: M61/Goldilocks (like STARKs) ✅ Pay-per-bit commitment costs (like Binius) ✅ Plausibly post-quantum secure 🔗 eprint.iacr.org/2025/294 🧵

The new UIDAI notification that allows private entities to use Aadhaar infrastructure for authentication could set up the marketplace of age verification service providers that we need. My latest Ex Machina article. exmachina.in/05/02/2025/new…

@nibnalin @daimo_eth Excited to see what you work on next. Your Nova Scotia repo helped me understand the potential of folding schemes. Thanks for that! I plan to dig into the Daimo codebase to understand how to seamlessly obtain user consent via signatures. Thanks again!

@daimo_eth What's next for me? I don't have any plans for now. I'm taking a step back to evaluate what excites me and explore without immediate goals. If you're working on interesting problems or deep ideas I'd love to chat.

Personal update: I’m stepping away from @daimo_eth

Snapchain testnet is now live. All Farcaster data is being streamed into Snapchain in parallel and things are working smoothly. A team of 3 engineers went from problem statement to working, novel blockchain design in 6 months. I think this might be a new record.

@wyatt_benno MIT license would be awesome. I hope people don't ditch it because the school hired Gary

@sarva_v Will look into this tomorrow. Likely MIT.

👨‍💻We implemented Nebula in our zkEngine (v0.9) (a WASM zkVM), it's open sourced w/ contributing to Microsoft/Nova repo. Running some tests today... with wasm files from the web and its 🔥🔥🔥 The NovaNet team is killing it! Cool features: 1. It's privacy preserving by default using updated HyperNova paper /w zk. 2. Choose the step size and it works almost anywhere. 500MB - 1GB prover~, works well in browser just like WASM. If you want to go crazy with a large machine you can! Just pick large step sizes. 3. SOTA for zkVM memory checks, pre-'Twist and Shout' 😆 We will do a separate post for this. 4. NIVC - (non-incremental verifiable computation) which has a few special traits like: stop and go proving, specialization circuits, more. Why another zkVM? Because we want ours to work in small spaces like browsers. == We have a few crazy cool use-cases we are dog-fooding 🐶 == Thanks to — all of the amazing open-source implementors, researchers, and educators. 🙏 What's Nebula again? eprint.iacr.org/2024/1605 Show me the code: github.com/ICME-Lab/zkEng…

@farcaster_xyz warpcast.com/sarva.eth

please reply thanks

Feb 1st is the application deadline for microgrants from @PublicGoodCrypt Research areas include ZK, FHE, MPC, PIR, PQC. Grant amount: 5k USD I am not affiliated. Just a friendly neighbourhood reminder uncle.

Use @NoirLang with the Barretenberg backend - faster proofs - works in browsers, even mobile* - no app-level trusted setup + small universal SRS download (~10mb vs 600mb for an equivalent groth16) - relatively expensive solidity verifier (but only $0.03 on L2 + optimizations otw)

Introducing MicroNova, a new variant of Nova that features an efficient on-chain verifier: eprint.iacr.org/2024/2099.pdf

Organizing a blockchain workshop at my alma mater IIT-Bombay on Dec 16. (w/ @sarva_v @vinayrib at @IITBTrustLab). link: trustlab.iitb.ac.in/event/scpp-blo… Exciting talks from John Augustine (IITM), @ChayaGanesh, Chethan Kamath (IITB), @pratyaycrypto & others at IITB.