Security Engineer – AI Model Protection & Mobile Security

We are building Chat application, an AI-powered communication app with proprietary STT, Machine Translation, and TTS models. We need an experienced AI/ML Security or Mobile Application Security Engineer to design and implement a defense-in-depth architecture to make our models significantly harder to extract, reverse-engineer, or reuse. Key Responsibilities::: • Encrypt and securely package AI models. • Design secure model loading and runtime execution. • Implement secure key provisioning and management. • Use Android Keystore/hardware-backed security where available. • Protect against APK extraction, reverse engineering, Frida/runtime hooking, debugging and memory dumping. • Implement model integrity and anti-tampering protections. • Secure model download/provisioning APIs. Implement authentication, authorization, rate limiting and device attestation. • Threat-model realistic model-theft attacks. • Conduct security testing and penetration testing. Important :::: We understand that on-device models cannot be made 100% impossible to extract. We specifically do not want• • • • AES-encrypted model + hard-coded key inside APK. We want a practical architecture that substantially increases the difficulty and cost of model theft. We are interested in evaluating:::: • On-device vs server inference • Encrypted models + secure runtime loading • Hardware-backed key storage • Server-controlled key provisioning • Device attestation • Model partitioning • Native C/C++ execution • Runtime protection • TEE/secure hardware where practical Required Experience ::: • Android security / NDK / JNI • C/C++ and ARM • Cryptography & key management • Reverse engineering • Frida, Ghidra, IDA Pro • Anti-tampering / anti-debugging • Secure API design • Threat modeling • ONNX / PyTorch / TFLite / Core ML or similar • • • Strong preference• • • candidates with previous experience protecting on-device AI/ML models. • • • • Deliverables• • • • • Threat model • Security architecture • Encrypted model packaging • Secure model loading • Key management/provisioning • Mobile/native implementation • Backend security • Anti-tampering protections • Security testing • Documentation