At Blockradar, we’ve developed a security framework that allows us and eventually our users to prove that private keys are never stored, that no single engineer can access or reconstruct private keys, and that the system remains safe even if all systems are compromised. At the highest level, Blockradar runs all critical workloads using real-time key computation and our custom Advanced AES-256-GCM encryption. We store only encrypted seed phrases (dual mainnet/testnet system) and derivation paths, never private keys. Every time a transaction needs to be signed, we compute the private key in real-time, use it once, and immediately destroy it. This helps to ensure that we can trust the secure applications running our infrastructure, which perform actions such as private key generation, transaction signing, and fund management. Our architecture gives us end-to-end transparency into the code we’re running and ensures no single developer at Blockradar can access private keys or reconstruct core secrets. We augment that trust in the application layer by ensuring the applications only act on verifiable data; every single change to a customer’s organization data must be approved and cryptographically stamped before it is considered by our systems. The system architecture ensures that even if all systems are compromised, your assets remains safe because private keys are never stored. Over time, we plan to open source more of this stack to enable other developers to verify our security claims themselves, and to allow clients to integrate with our security framework via their own applications. This, and more, is covered in depth in our documentation. Our security architecture is the foundation of the Blockradar product. Read on for deeper dives on each of the topics referenced above.