Pre-doctoral Position

Role Description

The IBM Research Zurich Laboratory and TU Wien have a fantastic opportunity for a highly motivated student to join our team as Doctoral Candidate (DC) working in a unique research-corporate environment to contribute to advancing the state-of-the-art in Reinforcement Learning (RL) techniques for code generation. The DC will have the opportunity to work with researchers advancing LLM code generation in open-weights enterprise models (IBM Granite model family, available in Hugging Face). The novel LLM methodologies incorporating RL will be applied to develop reliable data exploration tools that deliver accurate and informative responses to user queries while adhering to the confines of the domain knowledge.

Core Activities

During the doctoral studies, the DC will perform research activities in some of the following areas:

• Develop control methods for code models to ensure high-quality, domain-aligned responses from conversational interfaces for data exploration and analysis
• Leverage behaviour cloning, offline reinforcement learning and Reward Augmented Decoding to enhance and quantify output confidence, robustness, and domain alignment
• Design and advance domain specific synthetic data generation systems (SDG) to create instructions and verifiable datasets
• Design novel approaches to automate test case generations
• Contribute to the publication of research findings in leading AI conferences and journals
• Collaborate closely with researchers, engineers, and other teams within IBM Research and TU Wien
  

The candidate will be hired by the IBM Research Zurich Laboratory, having the opportunity to work in a unique corporate environment, acquire experience in several areas, publish in top international conferences, as well as deal with clients on real business cases. At TU Wien, the DC will be supervised by Prof. Katja Hose , in addition the candidate will also spend up to three months at Aarhus Universitet during a planned secondment.

Minimum Qualifications (mandatory)

• Outstanding university track record, with background in Computer Science, Artificial Intelligence, Statistics, or equivalent fields
• Proficiency in Python and experience with at least one deep learning framework (preferably PyTorch)
• Proficient working in Unix/Linux environments
• Team player, self-motivated with a passion for technology and innovation
• Ability to speak and write in English fluently
  
• Passion for code generation
  
  

Preferred Qualifications

• Experience in machine learning, deep learning, reinforcement learning (RL), large language models (LLMs)
• Experience in code generation, compilers, and code representation techniques
• Proficiency in Git and experience in common ML ops practices
• 3+ years of proved programming experience in Python (or equivalent C/C++ experience), plus familiarity with other languages (e.g., SQL, JavaScript)
• Experience with GPU clusters and job schedulers (e.g., Slurm)
• Independent worker with the ability to effectively operate with flexibility in a fast-paced, constantly evolving team environment
• Bonus: writing parsers, interpreters, or compilers and contributing to open-source projects
  
  

Conditions to apply

The researcher must not have resided or carried out their main activity (work, studies, etc.) in the country of the host organization (Switzerland) for more than 12 months in the 3 years immediately prior to the start date of the PhD.

This position is part of the European program ARMADA . If you are interested in this exciting position, please submit your most recent curriculum vitae and a short motivation letter. The interview process will include a programming interview and a ML/AI interview.

Diversity

IBM is committed to diversity at the workplace. With us you will find an open, multicultural environment. Excellent flexible working arrangements enable all genders to strike the desired balance between their professional development and their personal lives.

How to apply

Please submit your application through the link below.