Industrial Postdoc - AI Compilers in Köln

Roofline is building a software development kit to run any model on disruptive hardware at the edge.

We are looking for an Industrial Postdoc who wants to bridge cutting-edge research and real-world products.

You will turn our R&D results into publications, acquire and lead publicly funded research projects with academic and industry partners, and help shape our research agenda.

You will work alongside world-class engineers to enable the full spectrum of AI applications across heterogeneous hardware platforms, with the impact of a startup and the research depth of an institute.

Why us?

• The opportunity to grow. Be at the forefront of building and shaping an industry-defining venture. Shape our research strategy and see your ideas go from paper to product.
• The team to succeed. We are building the leading AI compiler team in Europe. Join our future-focused team to contribute to meaningful, disruptive innovations with us.
• The freedom to publish. You publish under your own name, and we support conference travel and academic collaborations, so your research career keeps moving forward.
• The flexibility you need. We want you to work in the most productive and sustainable way. Need flexible hours or additional time off to recharge? We’ve got you covered.
• The compensation that rewards.

We are looking for world-class talent and pay accordingly.

We offer equity for every employee since we truly want you to participate in our success.

• The culture you desire. We actively make space to build our culture. We bring you on site regularly and organize dedicated retreats for the whole team.
• Your mission
• Win funding for our next big ideas: spot the right national and European calls, build consortia, and coordinate and write grant proposals from first idea to submission
• Lead funded projects from kick-off to final report: track milestones, deliverables and budgets, and act as the main contact for funding agencies and consortium partners
• Turn our R&D results into high-quality papers for leading compiler, systems and ML conferences and journals, working closely with our engineers
• Shape our research agenda where science meets startup, and translate our technical roadmap into compelling research and funding strategies
• Build and grow partnerships with universities, research institutes and industry across Europe
• Represent Roofline in the research community at conferences, workshops and collaborations

Your profile

• Ph D in Computer Science, Electrical Engineering or a related field, or an MSc with substantial research experience
• Experience writing and coordinating research grant proposals, ideally at national or European level
• A publication record in compilers, ML systems, HPC, computer architecture or a related area
• Hands-on programming skills in Python or C++, or the ability to build working code with AI tools, to run experiments and get to publishable results
• Enough technical depth to understand our compiler technology and turn it into convincing proposals and papers
• Excellent written English and the ability to explain complex technical work to different audiences, from reviewers to customers
• Strong organizational skills and the ability to juggle several proposals, projects and deadlines at once
• An entrepreneurial mindset: you see opportunities, take ownership, and want to build something rather than just study it
• Nice to haves
• A track record of successful grant applications, or experience managing funded projects (e. g., Horizon Europe, EIC, BMFTR)
• An existing network in the European research community
• German language skills (helpful for national funding programs)
• Experience leading or coordinating research consortia
• Familiarity with MLIR or LLVM, AI accelerators, or hardware architectures for CPUs, GPUs or NPUs
• Experience benchmarking or profiling AI models on embedded hardware or So Cs

About us

At Roofline AI, we enable the edge-AI products you dream of.

Our mission is simple: To bring the power of chips into the hands of innovators, and to make it radically easier to use cutting-edge AI models on the best hardware.