LambdaClass

LambdaClass is a venture studio founded in 2014, specializing in building software that is fast and provably correct. They use Lean 4, Rust, and Erlang/Elixir to build systems where bugs are not acceptable, working across blockchain infrastructure, AI, finance, aerospace, and automotive industries.

Core Expertise

LambdaClass builds verified compilers, proof systems, distributed systems, and safety-critical software. Their code secures and moves billions in value across production systems. Founded by Federico Carrone, the company comprises 100+ engineers specializing in formal verification, cryptography, distributed systems, and high-performance software. LambdaClass is part of Ergodic Group, with teams in Buenos Aires, Montevideo, and Barcelona.

What They Do

Verified Optimization

LambdaClass proves optimizations correct in Lean 4, then generates production code from those proofs. The mathematical correctness is machine-verified.

Verifiable Computation

They build zero-knowledge proof systems and verifiable execution environments, enabling computation verification without re-execution.

High-Performance Systems

Using Rust, RISC-V, GPU-optimized code, and Erlang/Elixir, they build high-throughput backends with strict latency requirements, including execution clients, proof infrastructure, and custom VMs.

Safety-Critical Software

They develop software for aerospace, automotive, and finance—industries where bugs can be fatal or cost millions. Their software features machine-checked proofs of correctness.

AI Infrastructure

LambdaClass builds decentralized training networks, inference systems, and high-performance backends that enable AI training to scale beyond centralized clusters.

Post-Quantum Cryptography

Anticipating quantum computers breaking current cryptographic schemes, they're already building post-quantum clients and signature systems.

Notable Clients and Projects

Levenue: €1B in loans per year. Europe's largest revenue-based financing platform. LambdaClass rebuilt and runs the infrastructure.

Commit-Boost: Used by 40% of Ethereum validators. Open-source validator sidecar, grant-funded.

Nous Research: Building Psyche, a Rust-based decentralized AI training network enabling compute contribution over standard broadband.

Miden: 18+ months building the Miden client for transaction execution and proving on the Miden network.

Aligned: ZK security research, including discovery of critical SP1 exploit, and Ethrex integration for rollup infrastructure. Building LambdaVM, a RISC-V zkVM for Ethereum.

Starknet: Built the Rust Cairo VM powering Starknet's sequencer, achieving 30K to 220K Cairo steps per second.

Philosophy and Engagement

LambdaClass believes most software is built on faith rather than mathematical guarantees. As AI generates more code with less human review, they argue the need for formal verification is growing. They discovered and responsibly disclosed a critical exploit in Succinct's SP1 zkVM, demonstrating their security expertise.

Their engagement model involves becoming the engineering core for clients, fixing technical problems, rebuilding teams if needed, and serving as long-term partners. In many cases, they become the entire tech team. They work with companies where software correctness is a hard requirement—regulated industries, financial infrastructure, and systems where failure is measured in dollars or lives.

Patronage and Community Investment

LambdaClass invests in the tools and institutions they depend on:

• Universidad de Buenos Aires: $1M donation for scholarships in Computer Science, Mathematics, and Data Sciences
• Rust Foundation: $100K donation to support the language's long-term health
• Gleam and Roc Programming Languages: Monthly donations
• Argentine Astronomical Association: $10K/year covering IAU membership and journal sponsorship

Technology Stack

Their stack centers on three pillars:

• Lean 4 for formal verification and machine-checked proofs
• Rust for performance-critical systems with compile-time guarantees
• Erlang/Elixir for fault-isolated systems with hot code upgrades and massive concurrency