PhD in AI-assisted reverse engineering of integrated circuits (1 FTE)

Tilburg University; Tilburg; On-site; EUR 30,000 - 42,000; Full time

Job summary

Tilburg University’s Cognitive Science & AI Research Center invites applications for a PhD researcher position within the SHIELD project, an AI-assisted reverse engineering initiative for hardware assurance and intelligence analysis under the Department of Intelligent Systems. The role sits at the intersection of computer vision, micro-electronics analysis, and hardware security.

You will develop a pipeline for analyzing high-resolution microscopy data of ICs, combining SEM, PEM, and FIB imaging

Qualifications

• PhD in computer vision, AI, electrical engineering, or a related field.
• Strong experience with deep learning, image analysis, and multimodal data.
• Ability to collaborate across academic and defence partners.

Responsibilities

• Develop a pipeline for multilayer segmentation and nanoscale analysis from microscopy data.
• Design graph-based models to reconstruct higher-level logical blocks and detect anomalies.
• Evaluate memory extraction and vulnerability analysis using physically grounded image data.

Skills

Deep learning

Computer vision

Hardware security

Microscopy analysis

Education

PhD or equivalent in a related field

Tools

SEM

FIB

PyTorch

Graph-based inference

Job description

The Cognitive Science & Artificial Intelligence Research Center at Tilburg University hosts the Secure Hardware Intelligence Extraction and Logic Deconstruction (SHIELD) research project, a scientific initiative focused on AI-assisted reverse engineering of integrated circuits for hardware assurance and intelligence analysis. The project is conducted within the Deep Learning for Perception and Data Science, Safety and Security research units, in collaboration with the Data Science Centre of Excellence of the Netherlands Defence Academy and the Netherlands Forensic Institute. The PhD researcher will operate at the intersection of computer vision, micro-electronics analysis, and hardware security, and will work under the supervision of researchers within the Department of Intelligent Systems. The PhD researcher will be formally embedded in the Department of Intelligent Systems (DIS) and will collaborate closely with the Department of Computational Cognitive Science (DCS). The SHIELD project is a collaboration between Tilburg University, the Data Science Centre of Excellence of the Netherlands Defence Academy, and the Netherlands Forensic Institute, combining expertise in artificial intelligence, micro-electronics analysis, and hardware security. The project focuses on developing AI-assisted methodologies for reverse engineering integrated circuits in order to support hardware assurance, threat analysis, and intelligence extraction workflows. The Tilburg PhD researcher will develop a methodological pipeline for analysing high-resolution microscopy data of integrated circuits, integrating computer vision, deep learning, and logical reconstruction techniques. The research investigates how multimodal imaging modalities - including scanning electron microscopy (SEM), photon emission microscopy (PEM), and focused ion beam (FIB) imaging - can be combined with AI to reconstruct nanoscale chip structures and infer functional behaviour from physical layouts. The project addresses the challenge of extracting reliable structural and logical information from complex semiconductor architectures under realistic imaging constraints. Specifically, the researcher will focus on the following three main tasks:(a) Developing deep learning pipelines for multilayer segmentation and nanoscale transistor classification from microscopy images; (b) Designing graph-based inference models capable of reconstructing higher-level logical blocks and identifying anomalous structures indicative of Hardware Trojans; (c) Evaluating methodologies for memory extraction, vulnerability analysis, and logical reconstruction using physically grounded image data. The research explores how AI-driven analysis can move beyond manual reverse-engineering workflows by automating feature extraction and structural interpretation while remaining robust to noise, obfuscation, and architectural variability. Because the work operates at the intersection of scientific research and hardware security, the project carefully balances methodological innovation with responsible handling of sensitive datasets and forensic constraints. The overall goal is to develop a reproducible proof‑of‑concept methodology that enables scalable analysis of integrated circuits for both defensive hardware verification and intelligence‑oriented analysis of captured electronics. The PhD researcher will evaluate the scientific validity and limitations of the proposed methods and contribute to open-access publications and technical documentation. The SHIELD project adopts a forward‑looking research perspective aimed at strengthening long-term capabilities in trusted hardware, semiconductor sovereignty, and cyber‑electromagnetic resilience. The PhD researcher will primarily work within Tilburg University’s AI research infrastructure, focusing on algorithm development, model training, and methodological validation. Experimental validation and imaging data originate from collaborations with specialised forensic and defence partners. In particular, the Netherlands Forensic Institute (NFI) advises on high-resolution microscopy datasets and domain expertise in semiconductor analysis, including techniques such as scanning electron microscopy and focused ion beam imaging. These facilities enable realistic evaluation of reverse‑engineering methods under practical constraints while ensuring that the research remains grounded in real‑world forensic workflows. Connection to external stakeholders: The SHIELD project is conducted in collaboration with the Data Science Centre of Excellence of the Netherlands Defence Academy and additional defence stakeholders who provide operational context and feedback during milestone evaluations. Their role is advisory and focused on validating methodological relevance rather than directing the scientific outcomes. This collaboration ensures that the research addresses real hardware assurance challenges while maintaining academic independence and open scientific dissemination.

• Developing deep learning pipelines for multilayer segmentation and nanoscale transistor classification from microscopy images;
• Designing graph-based inference models capable of reconstructing higher-level logical blocks and identifying anomalous structures indicative of Hardware Trojans;
• Evaluating methodologies for memory extraction, vulnerability analysis, and logical reconstruction using physically grounded image data.