Bell Labs Internship on Encrypted and Verifiable ML Inference on Edge Devices (PhD)

Job Description:

Fully homomorphic encryption (FHE)-based encrypted machine-learning inference has attracted increasing attention. Conducting such inference on resource-constrained edge devices introduces a distinct set of challenges. In this project you will first evaluate machine-learning models that are amenable to efficient FHE deployment, and subsequently assess the feasibility of realizing the selected model using a verifiable FHE scheme.

* You are expected to familiarize yourself with a range of edge-optimized machine-learning models FHE schemes.
* You will critically assess the advantages and disadvantages of various models and FHE schemes, ultimately selecting a few promising combinations for further development.
* You will develop functional prototypes of machine-learning inference using the chosen models and encryption schemes, and evaluate their performance on resource-constrained edge devices.
* Consolidate your research in a scientific publication

Candidate Requirements:

Qualifications
* Student enrolled in PhD Computer Science/Engineering.
* Familiarity with Fully Homomorphic Encryption techniques and frameworks
* Familiarity with Machine Learning techniques and frameworks
* Good programming skills in Rust, C, or C++ and in Python or Golang
* Language skills: English

Location: Stuttgart (Germany)