Postdoctoral Research Visit in Embedded Machine Learning, TinyML, and Privacy-Preserving Edge AI Inria in France

Inria is advertising a postdoctoral research visit in Embedded Machine Learning, TinyML, privacy-preserving machine learning, and edge AI. The position is based at Inria Saclay with time also spent in Berlin at Freie Universität Berlin, reflecting a collaboration between the two sites.

The research topic combines two complementary directions: TinyML on constrained microcontrollers and privacy-preserving split computing for neural network inference. The postdoc will work on efficient AI model execution on embedded hardware such as Nordic nRF52, STM32, ESP32, or RISC-V platforms, with networking technologies like BLE, 802.15.4, or LTE-M. Prototypes may be developed with Ariel OS or RIOT OS, and the project builds on prior work such as Ariel-ML and RIOT-ML.

On the privacy side, the role focuses on split inference, reconstruction risks, and defense mechanisms that reduce leakage while preserving performance. The work includes experimental prototypes, evaluation on heterogeneous embedded hardware, and interaction with researchers and engineers in machine learning, PETs, secure low-power IoT protocols, and open-source communities.

Eligibility: a PhD or equivalent is required. Strong experience in machine learning, MLOps, privacy-enhancing technologies, embedded Rust and/or C, Python, low-level optimization, RTOS or bare-metal development, and network protocol stacks is preferred. Good scientific English and the ability to work in distributed teams are expected.

Funding and conditions: this is a fixed-term postdoctoral contract for 2 years, renewable. The gross monthly salary is 2,788 euros. Benefits include subsidized meals, partial public transport reimbursement, 7 weeks of annual leave plus 10 RTT days, teleworking options, equipment support, training access, and social security coverage.

Deadline: apply by 2026-08-31. Applications must be submitted online through the Inria recruitment website.