Alice Liu: Welcome Vanishree, thank you for joining us today.
Vanishree Rao: Thanks, Alice. It’s great to be here.
AL: Let's start off by introducing yourself. Could you share your journey from cryptography to web3, and what led you to focus on zero-knowledge proofs?
VR: Absolutely. Currently, I'm the CEO of Fermah, which is a universal proof generation layer designed to generate proofs efficiently for any proof system. We’re building a marketplace where machines like FPGAs and TPUs supply computation for generating proofs across any zero-knowledge use case. We’re already on devnet and permissioned mainnet, and it's been an incredible journey so far.
My story actually started about 20 years ago when I first learned about zero-knowledge proofs during my undergrad in India. My professor introduced me to the concept, and it immediately struck me as something amazing. Imagine convincing someone of the truth of a statement without sharing any underlying information - how could that even be possible? The paradox intrigued me deeply.
I spent months studying cryptography and number theory from a book by Neal Koblitz, which reads almost like a novel, and it truly sparked my interest in ZK. After undergrad, I moved to the U.S. to pursue a Ph.D. in cryptography at UCLA, working under Amit Sahai, who’s done foundational work in zero-knowledge proofs and multiparty computation.
Later, I joined the Mina Protocol early on, working on consensus and the zk-SNARK implementations. The experience gave me a better understanding of the infrastructure needed to apply ZK practically. Ultimately, I realized there was a fundamental gap in scalable infrastructure for ZK proofs - a lack of accessible proof generation solutions - and that's what motivated me to start Fermah last year.
AL: That’s quite the journey, from studying the theory to applying it directly in practice. With all that experience, how do you see the evolution of the ZK space from when you started to now?
VR: Good question. I think the evolution of the ZK space has been fascinating. There have been several pivotal phases. Initially, ZK was largely confined to academic research, with a lot of focus on specific protocols for targeted use cases - for example, proving ownership of secrets in cryptographic protocols like HTTPS. Then came the generic protocols that could handle any computation, shifting from interactive to non-interactive proof systems, like zk-SNARKs, which were a breakthrough for practical implementation.
After that, we saw a surge of interest from the crypto industry, which drove the development of more efficient and standardized proof systems, like the one used in Zcash. Over time, the focus shifted from specific ZK circuits to zkVMs, which allow developers to work with generalized virtual machines, dramatically lowering the barrier to entry.
Looking ahead, I see a renewed interest in specialized ZK circuits. For mature projects, the initial constraints - like time to market or engineering costs - start to diminish, and performance becomes the priority. At that point, building custom circuits instead of relying on zkVMs could lead to significant efficiency gains.
AL: So, it sounds like there is always this trade-off between specialization and generalization, depending on the project's maturity. Beyond this, what do you see as the most pressing challenges in the ZK space today?
VR: I’d say the key challenges are around proving infrastructure. The ZK adoption process has multiple layers: identifying the business logic to be proven, selecting the right proof system, setting up suitable hardware, and orchestrating the entire proof generation and settlement process.
Most layers have seen substantial progress, but the infrastructure layer still has gaps. That's why we’re building Fermah—a universal proof generation layer that abstracts away the complexity of setting up and running proving infrastructure. We’re focusing on issues like optimizing incentive mechanisms to ensure economic feasibility and reliability for proof generation without sacrificing performance.
Another challenge is data privacy when proving computations involving sensitive information. This is why we’re working on data-protected proving delegation, which combines zero-knowledge proofs, multi-party computation, and trusted execution environments to allow secure proving without exposing sensitive data.
AL: That's really insightful. I also noticed you've been discussing some fascinating applications for ZK beyond blockchain, like in investigative journalism and audits. Can you elaborate on those?
VR: Sure. Outside of blockchain, one interesting application is in investigative journalism. Zero-knowledge proofs can provide proof of provenance for digital media, ensuring that photos or videos are authentic and haven’t been tampered with—which is especially important given how easy it is to manipulate images today.
Another promising area is corporate audits. Auditing large corporations is expensive and time-consuming, but ZK can help make the process efficient, cost-effective, and less error-prone. By generating cryptographic proofs, companies could prove compliance or solvency without having to reveal sensitive transactional data. This is a prime example of where ZK could have an impact beyond web3.
AL: It's exciting to think of ZK becoming even bigger than blockchain, given how much broader its potential applications are. Do you see ZK as the dominant technology for cryptographic integrity, or are there other cryptographic primitives that you're also excited about?
VR: I think ZK is certainly one of the core tools for proving integrity, but it’s not a one-size-fits-all solution. Other cryptographic primitives like multi-party computation (MPC) and trusted execution environments (TEEs) also play a critical role. Each has its unique strengths - for instance, TEEs can help speed up MPC protocols, which tend to have high communication overhead.
These primitives can be used together to solve problems that none could solve alone. For example, if you need to prove a computation that involves sensitive data on a low-capacity device, you could split the data securely using MPC, process it across multiple devices, and use ZK to generate a proof while ensuring privacy. They’re complimentary, not competitors.
AL: That’s a great perspective, and I think your analogy of tools - like comparing a knife and a blender - sums it up well. Lastly, could you share more about the design considerations you prioritize at Fermah, particularly with respect to proving networks?
VR: Our main focus is on two things. First, getting the mechanism design right - optimizing proof generation in terms of cost and speed. Our orchestration should add minimal overhead while maintaining optimal performance. Second, building data-protected proving delegation, which allows us to work with both privacy-sensitive and privacy-agnostic computations. This approach will help us serve a broader set of use cases across the ZK ecosystem.
Ultimately, our goal at Fermah is to make ZK adoption as seamless as possible. We want developers to interact with a simple interface where they define constraints, choose their preferred proof system, and let the infrastructure handle the rest—all while keeping costs low and performance high.
AL: It sounds like an exciting future ahead for Fermah and for ZK in general. Thank you so much for sharing your insights with us today, Vanishree. Where can people learn more about you and Fermah?
VR: Thanks, Alice. The best way to connect is on Twitter (@fermah_xyz) or through our website at fermah.xyz. We’re always eager to collaborate, and we believe there’s a lot of room for collaboration as the ZK space grows.
AL: Wonderful. Thank you again, Vanishree, for your time. It’s been a pleasure having you on House of ZK Radio.
VR: Thank you, Alice. It was a pleasure.
To learn more about Fermah, visit fermah.xyz/ or @fermah_xyz on X.
You can also follow @vanishree_rao for regular updates on developments at Fermah.
