SecretBit Ventures builds trustworthy AI systems grounded in mathematics, cryptography, and privacy. Founded by Dr. Sameer Wagh, we work at the intersection of agentic AI, secure computation, and privacy-enhancing technologies — turning cutting-edge research into production systems.
What We Do
- Private AI & Secure Computation: Privacy-preserving machine learning, secure multi-party computation, homomorphic encryption, and federated learning — from protocol design to production deployment
- Agentic AI Infrastructure: Secure enclaves for AI evaluation, confidential cross-organization model testing, LLM deployment optimization, and AI-driven code analysis
- Mathematical Foundations: Rigorous cryptographic protocol design, novel proof techniques, and formal security guarantees rooted in deep mathematical expertise
Selected Projects
Secure Enclaves for AI Evaluation
Anthropic & UK AI Safety Institute
Designed and deployed the first pilot of secure enclaves for AI evaluation, enabling confidential cross-organization model testing and setting precedent for industry adoption.
Private Network Trace Analysis
University of Virginia
Privacy-preserving analysis of sensitive network trace data, enabling secure collaboration on network security research without exposing raw traffic.
Efficient MPC via Wavelet Transforms
Nillion Inc.
Novel wavelet transform-based protocols for efficient secure multi-party computation, pushing the boundaries of practical MPC performance.
Automated AI Code Analysis
Devron Corporation
Built an AI system for automated code analysis at a fraction of the cost, cutting monthly LLM endpoint costs by over 90% through architectural and deployment optimizations.