Senior AI Researcher- Reinforcement learning (f/m/d)

Aleph Alpha

OUR MISSION Aleph Alpha is one of the few companies in Europe doing serious foundation model pre- and post-training. We're building models that have general-purpose capabilities, and specifically excel at addressing the needs of our customers. We're looking for exceptional Software Engineers to join our model training team. Most of the team is based in Heidelberg . TEAM CULTURE At Aleph Alpha, we foster a culture built on ownership, autonomy, and empowerment. Teams and individual contributors are trusted to take responsibility for their work and drive meaningful impact. We maintain a flat organizational structure with efficient, supportive management that enables quick decision-making, open communication, and a strong sense of shared purpose. We believe a strong engineering culture is the key to model training success. We like Extreme Programming and favor trunk-based development. We often mob-program, which keeps us aligned and means we always learn from each other. ABOUT THE ROLE As a Senior AI R&D Engineer in Model Training (f/m/d), you will join our model training team in Heidelberg or Berlin. Most of the team is based in Heidelberg. The model training team is training models, but more importantly in order to do so building and improving our model training pipeline. * You will work across our full stack. Some weeks you might be optimizing how training loads are scheduled on our cluster and making the pipeline more robust and performant so we can iterate faster. Other weeks, you will be enabling large-scale code execution for reinforcement learning. And at other times, you might dig deep into our evaluation codebase to lift inference throughput on evals. * No two days are the same. Things move fast, and your ability to focus and prioritize is what lets you unblock the team day-to-day while designing the high-quality tooling and infrastructure that speeds us up long-term. * We're still building out our training pipeline and infrastructure. Some pieces exist, some do not, and you will have real influence on what gets built and how. Your work directly shapes how quickly we can experiment and improve our models. * Identifying issues and taking initiative is what drives our day-to-day work. * Working mode, design decisions, priorities are all done in your team together with your team mates. The strongest teams often use highly collaborative techniques, like mob or pair programming. * Due to the highly collaborative nature of our work you are expected to be on-site at least two days a week. If you are in Berlin, you should expect to travel around every second month for a few days to Heidelberg. * We are specifically looking to strengthen the Software Engineering in the teams which are research heavy. Major points include improving the maintainability of our code through tests and design & architecture changes. REQUIREMENTS * At its core, you need to be able to dive into a codebase, change the behavior as necessary, and after the change the code should be better tested and easier to understand than before. * As such we look for candidates for an understanding of how to design for low coupling and high cohesion paired with the ability to incrementally refactor. * Experience with the following technologies is likely to be helpful: * Test Driven Development (TDD) * Trunk based development * Continuous integration * Domain Driven Design NICE-TO-HAVES * Most of our code is in Python, there is also the occasional Rust codebase. * Large parts of our training pipeline are built on top of Kubernetes. * We do not hire for skills in a particular technology or language, but rather ability to learn and mindset, yet experience with these technologies is a plus. WHAT WE OFFER * Become part of an AI revolution! * 30 days of paid vacation * Access to a variety of fitness & wellness offerings via Wellhub * Mental health support through nilo.health * Substantially subsidized company pension plan for your future security * Subsidized Germany-wide transportation ticket * Budget for additional technical equipment * Flexible working hours for better work-life balance and hybrid working model * JobRad® Bike Lease

Aleph Alpha

OUR MISSION Aleph Alpha is one of the few companies in Europe doing serious foundation model pre-training. Our customers - in finance, manufacturing, public administration - need models that understand German, meet European regulatory requirements, and work reliably in high-stakes settings. We're building that in Heidelberg. We're growing our pre-training team and hiring someone to passionately work on data: defining what goes into our models, building the systems that source and prepare it, and ensuring our training team has the highest-quality data to push model capabilities forward. Team Culture At Aleph Alpha, we foster a culture built on ownership, autonomy, and empowerment. Teams and individual contributors are trusted to take responsibility for their work and drive meaningful impact. We maintain a flat organisational structure with efficient, supportive management that enables quick decision‑making, open communication, and a strong sense of shared purpose. About the role As a Senior AI Researcher for Pre-training Data, you will shape and improve the underlying scientific methodology behind our pre-training corpora while also co-engineering the software and systems that enable this. Working with engineers and other researchers to build scalable pipelines, you will focus on relevant theoretical and empirical research required to understand which data makes models perform best on our targeted capabilities. This role is for you if you have a strong background in large-scale language modeling and the scientific drive to answer complex questions about data scaling laws, synthetic data generation, and curriculum learning. In your day-to-day, you will design targeted ablations across various scales, derive and test hypotheses from training dynamics, develop novel algorithms for estimating data quality and performing data curation, and contribute to a range of engineering tasks which facilitate these research directions. Together with a collaborative team of engineers and researchers, you will have a direct impact on the fundamental knowledge and capabilities of the models we ship. You will also help or lead the writing of technical reports for internal and external readers, as well as presenting at and contributing to technical meetings and conferences on an as-needed basis. YOUR RESPONSIBILITIES * Innovation in Data-Centric AI: Stay at the bleeding edge of foundation model research. You will identify, implement, and iterate on novel approaches to estimating data quality, synthetic data generation, curriculum learning, and advanced curation techniques. * Data-to-Performance Science: Design and lead rigorous ablation studies across various scales. You will systematically analyse how changes in data composition, deduplication strategies, heuristic and model-based curation, and scaling laws affect training dynamics and target model and system capabilities. * Develop Novel Quality Signals: Move beyond basic perplexity filtering. Research and build advanced algorithms to score and select data, such as influence functions, gradient-based matching, or using smaller models to curate data for larger ones. * Cross-Functional Collaboration: Partner closely with a diverse team to scale your research from prototypes to trillions-of-tokens-scale pipelines, and work with the post-training team to ensure pre-training distributions effectively support targeted fine-tuning and customer-alignment. YOUR PROFILE Basic Qualifications * A deep understanding of machine learning theory, specifically regarding foundation model training dynamics, scaling laws, and data-centric AI. * Experience designing and evaluating complex ML experiments related to data composition, curriculum learning, or data quality on language model training. * Familiarity with statistical methods for evaluation and experiment design. * Ability to reason about the information-theoretic properties of a dataset and its predictive power for evaluated tasks: not just processing data, but understanding its signal. * Strong Python skills and comfort with ML tooling and deep learning frameworks (especially PyTorch). * Willingness to relocate to Heidelberg or travel at least fortnightly. Preferred Qualifications * PhD in machine learning, NLP, or equivalent research experience focusing on large-scale language modeling or data curation. * A history of contributions to top-tier venues (NeurIPS, ICML, ICLR, ACL, etc.) specifically regarding data curation, scaling laws, synthetic data, or LLM pre-training. * Experience training foundation models from scratch and diagnosing data-induced training pathologies. * Bonus, but not required: German language proficiency can be helpful for curating and assessing German-language data. COMPENSATION AND BENEFITS * Become part of an AI revolution! * 30 days of paid vacation * Access to a variety of fitness & wellness offerings via Wellhub * Mental health support through nilo.health * Substantially subsidized company pension plan for your future security * Subsidized Germany-wide transportation ticket * Budget for additional technical equipment * Flexible working hours for better work-life balance and hybrid working model * JobRad® Bike Lease

Aleph Alpha

OUR MISSION Aleph Alpha is one of the few companies in Europe doing serious foundation model pre-training. Our customers — in finance, manufacturing, and public administration — need models that understand German, meet European regulatory requirements, and work reliably in high-stakes settings. We’re building that in Heidelberg. We are hiring a Senior AI Researcher to join our Pre-training team and to advance the architecture and training of our next generation of foundation models. If you are excited about designing inference-efficient architectures, optimising training recipes that scale reliably, and training models on a large scale cluster (thousands of NVIDIA Blackwell GPUs), we would love to hear from you. TEAM CULTURE We foster a culture built on ownership, autonomy, and empowerment. Teams and individual contributors are trusted to take responsibility for their work and drive meaningful impact. We maintain a flat organisational structure with efficient, supportive management that enables quick decision-making, open communication, and a strong sense of shared purpose. We collaborate closely on complex technical problems, working in pairs or using mob programming to resolve challenging issues. ABOUT THE ROLE As a Senior AI Researcher in Pre-training (f/m/d), you will own the critical technical levers that determine the success of our next-generation models: architecture, optimization, stability, and scaling. Working at the high-leverage intersection of research and engineering, you will translate mathematical reasoning and empirical observations into principled training decisions - from small-scale proxy experiments to multi-thousand-GPU runs. We are looking for an expert who can combine rigorous experimental design with high-quality production code, directly influencing model quality, run reliability, and the efficiency of the models we ship. YOUR RESPONSIBILITIES * Recipe & Architecture Optimization: Own core elements of the training recipe (optimizers, schedules, initialization) and design PyTorch-based architectural improvements to maximize convergence, stability, and training efficiency. * Scaling Strategy & Predictability: Develop hyperparameter scaling laws and scale-up methodologies, using small-scale proxy experiments to reliably predict multi-thousand-GPU behavior and de-risk major training decisions. * Stability, Diagnostics & Debugging: Investigate complex convergence issues (loss spikes, divergence) and resolve hard-to-reproduce distributed system failures like communication bottlenecks, race conditions, and synchronization errors. * System-Model Co-Design: Partner with Compute Performance, Data, Evaluation, and Post-Training teams to align the model lifecycle with hardware constraints, memory bandwidth, and communication topologies. CORE QUALIFICATIONS * You are proficient in Python and deeply familiar with PyTorch-based training workflows. * You have a strong track record in machine learning research and software engineering, demonstrated through shipped models, impactful open-source contributions, or published research. * You have a strong mathematical foundation and are comfortable reasoning formally about optimisation, scaling behaviour, and training dynamics. * You deeply understand transformer training dynamics, optimisation, and the behaviour of large distributed training jobs. * You can design rigorous experiments, reason clearly from noisy results, and translate empirical observations into robust training decisions. * Hands-on experience pre-training large models (e.g., 7B+ parameters) on substantial infrastructure (e.g., 100+ GPU clusters). * You apply strong software engineering practices, including writing maintainable, well-tested code and supporting reproducible experimentation workflows. * You are able to implement complex model architectures efficiently and reliably and to debug complex issues across model code, training dynamics, and distributed systems. * You collaborate effectively within a research and engineering team and communicate clearly about your work across Pre-training and the broader AAR/AA organization. * You are able to work in Germany and collaborate regularly on site in Heidelberg as part of the Pre-training team. PREFERRED QUALIFICATIONS (We encourage you to apply even if you don't check every box!) * Large-Scale Training: Hands-on experience training LLMs or multimodal models on large GPU clusters using distributed frameworks (e.g., Megatron-LM, DeepSpeed, torchtitan). * Predictive Scaling: Familiarity with scaling laws, hyperparameter transfer, or methods for predicting large-scale training behavior from smaller proxy runs. * Stability & Performance: Experience profiling distributed jobs and diagnosing training anomalies like loss spikes, numerical instability, or optimizer pathologies. * Advanced Architectures: Exposure to sparse training approaches (e.g., Mixture-of-Experts) and an understanding of their routing and systems trade-offs. * Track Record of Impact: Demonstrated research excellence through top-tier publications (NeurIPS, ICML, ICLR), impactful open-source contributions, or significant shipped technical work. * Systems Curiosity: Low-level kernel optimization is not required, but we highly value a strong curiosity about the hardware and systems constraints that shape scale. What we offer * Become part of an AI revolution! * 30 days of paid vacation * Access to a variety of fitness & wellness offerings via Wellhub * Mental health support through nilo.health * Substantially subsidized company pension plan for your future security * Subsidized Germany-wide transportation ticket * Budget for additional technical equipment * Flexible working hours for better work-life balance and hybrid working model * JobRad® Bike Lease