PhD Position in World Models for Software Architecture Synthesis
Develop formal world models to predict architectural consequences and synthesize software compositions using machine learning and formal methods.
Best for:Best for computer science or mathematics graduates with expertise in programming language theory, compilers, or model-driven engineering, who are passionate about formal methods and empirical validation of software systems.
Overview
This PhD position at Constructor University Bremen, in collaboration with Constructor Technology, focuses on advancing software engineering through the creation of world models for architectural synthesis. The project, titled Constructor Fabric, aims to transform informal business knowledge into production software by predicting outcomes of architectural decisions—such as cost, unintended side effects, and feasibility—using a formal architectural skeleton and domain-specific language. The successful candidate will contribute to designing and implementing a world model that integrates machine learning, formal methods, and software engineering to enable automated synthesis of software architectures. The work spans formal language design, architecture synthesis, and data-driven modeling, with a strong emphasis on empirical validation and falsifiability of assumptions.
What you'll do
- Design and implement a formal architectural skeleton for software systems, including gear contracts and typed attributed hypergraphs.
- Develop algorithms for architecture synthesis, treating it as an inverse problem to find optimal gear compositions given specifications and constraints.
- Contribute to the design of a domain-specific language (DSL) for software architecture, ensuring its static semantics and expressiveness.
- Build local transition models and pipeline surrogates to predict outcomes of architectural decisions with calibrated uncertainty.
- Reconstruct and analyze (specification → architecture) pairs from open repositories to validate and improve the world model.
- Collaborate on the development of a projectional editor for the DSL, leveraging JetBrains MPS for graph-based representations.
- Conduct empirical experiments to falsify assumptions and validate the model’s predictions against real-world telemetry.
Requirements
Degree required: Master of Science (MSc) or equivalent in computer science, mathematics, or a related mathematical natural science
Degree offered: PhD
Language: English; excellent proficiency required for collaboration and documentation
Instruction: English
Publications: A maintained open-source repository, peer-reviewed publication, or strong competition result demonstrating research completion
- Proven coursework in compilers, programming language theory, language design, or model-driven engineering
- Experience with formal methods or mathematical modeling of software systems
- Strong programming skills and ability to maintain open-source repositories
- Publication or competition record demonstrating research completion
- Fluent technical English for collaboration and documentation
- No nationality restrictions
- Must hold or be close to completing a Master’s degree in computer science, mathematics, or related field
- No age restrictions mentioned
- No prior experience caps, but evidence of research completion is strongly preferred
Funding & terms
How to apply
Deadline: 2026-10-30 (Europe/Berlin)
Applicants must submit a CV, cover letter (1–2 pages), a link to written code, a one-page research statement addressing a project question, Master’s and Bachelor’s degree certificates, and contact information for 2–3 academic referees. Applications are reviewed on a rolling basis, and shortlisted candidates will be invited for interviews.
- CV including publications, code, or research projects
- Cover letter (1–2 pages) explaining motivation and research fit
- Link to code written by the applicant
- One-page research statement addressing a project question
- Master’s and Bachelor’s degree certificates and transcripts
- Contact information for 2–3 academic referees
Contact
Constructor Knowledge Labs· Recruitment Contact
Create a free profile to unlock your personal match score, the full requirements, funding details and how to apply — in English or Persian.