Senior LLM Engineer | Relocation Offered

Multiverse Computing; Barcelona; On-site; EUR 110,000 - 160,000; Full time

Job summary

Multiverse Computing is seeking a Senior LLM Engineer to join our Research team in Barcelona. You will oversee the full model lifecycle: data handling, distributed training, post-training, compression, evaluation and efficient inference.

Our work includes compressing and healing open-weight models, SFT, RL and distillation across multimodal data, and building our own model family. You will collaborate with world‑leading researchers to publish papers and release code when appropriate.

Qualifications

• Master’s or PhD in Computer Science, AI, Mathematics or Physics, or equivalent industry experience.
• Senior: 4+ years of hands-on experience training or post-training transformer models.
• Strong PyTorch; comfortable with model code and parallelism.
• Experience with distributed training on multi-node clusters with Slurm/K8s.
• Hands-on post-training experience: SFT, DPO, and at least one RL method (PPO, GRPO or similar).
• Solid mathematical foundations in linear algebra, optimisation and probability.
• Good understanding of GPU fundamentals; memory hierarchy, mixed precision, profiling.
• Fluent English.

Responsibilities

• Train and post-train LLMs and multimodal models on multi-node GPU clusters using SFT, preference optimisation, RL (e.g. GRPO), and on-policy and offline distillation.
• Design compression techniques (quantization, pruning, quantization-aware training, quantum-inspired tensor-network methods) and healing recipes that recover the quality of compressed models.
• Build RL environments, reward functions and verifiers for reasoning, coding and agentic tasks, and extend RL methods to world models and Physical AI.
• Research and prototype post-transformer and hybrid architectures (state-space models, linear attention, MoE), from small-scale ablations to scaling recipes.
• Build data pipelines for curation, filtering, deduplication and large-scale synthetic data generation.
• Design rigorous ablations and evaluations on current benchmark suites (reasoning, coding, long-context, instruction following, agentic) as well as internal evals.
• Port and optimise models for inference on various accelerators and edge chips, profiling latency, memory and throughput.
• Write up experiments and results as internal reports and, where relevant, papers and open-source releases.
• Review code, mentor teammates, and follow and bring in the latest research.

Skills

PyTorch

Distributed training

SFT

RL methods

Quantization

Model compression

Education

Master's or PhD in CS/AI/Math/Physics

Equivalent industry experience

Job description

About Multiverse Computing

Multiverse is a well-funded, fast-growing deep-tech company founded in 2019. We are the largest quantum software company in the EU and have been recognized by CB Insights (2023 and 2025) as one of the 100 most promising AI companies in the world.

With 180+ employees and growing, our team is fully multicultural and international. We deliver hyper-efficient software for companies seeking a competitive edge through quantum computing and artificial intelligence.
Our flagship products, CompactifAI and Singularity, address critical needs across various industries:

• CompactifAI is a groundbreaking compression tool for foundational AI models based on Tensor Networks. It enables the compression of large AI systems—such as language models—to make them significantly more efficient and portable.
• Singularity is a quantum- and quantum-inspired optimization platform used by blue-chip companies to solve complex problems in finance, energy, manufacturing, and beyond. It integrates seamlessly with existing systems and delivers immediate performance gains on classical and quantum hardware.

You’ll be working alongside world-leading experts to develop solutions that tackle real-world challenges. We’re looking for passionate individuals eager to grow in an ethics-driven environment that values sustainability and diversity.

We’re committed to building a truly inclusive culture—come and join us.

Position

We are looking for a Senior LLM Engineer to join our Research team. You will work across the full model lifecycle: data, distributed training, post-training, compression, evaluation and efficient inference. Our past work includes compressing and healing open-weight models, post-training with SFT, RL and distillation, multimodal training, knowledge editing, and building our own family of models. Priorities change quickly in research, so we value strong fundamentals over narrow specialisation. When it makes sense, we publish papers and release models and code.

As an LLM Engineer, you will

• Train and post-train LLMs and multimodal models on multi-node GPU clusters using SFT, preference optimisation, RL (e.g. GRPO), and on-policy and offline distillation.
• Design compression techniques (quantization, pruning, quantization-aware training, quantum-inspired tensor-network methods) and healing recipes that recover the quality of compressed models.
• Build RL environments, reward functions and verifiers for reasoning, coding and agentic tasks, and extend RL methods to world models and Physical AI.
• Research and prototype post-transformer and hybrid architectures (state-space models, linear attention, MoE), from small-scale ablations to scaling recipes.
• Build data pipelines for curation, filtering, deduplication and large-scale synthetic data generation.
• Design rigorous ablations and evaluations on current benchmark suites (reasoning, coding, long-context, instruction following, agentic) as well as internal evals.
• Port and optimise models for inference on various accelerators and edge chips, profiling latency, memory and throughput.
• Write up experiments and results as internal reports and, where relevant, papers and open-source releases.
• Review code, mentor teammates, and follow and bring in the latest research.

Your qualification

• Master’s or PhD in Computer Science, AI, Mathematics, Physics or a related field, or equivalent industry experience.
• Senior: 4+ years of hands-on experience training or post-training transformer models.
• Strong PyTorch. You are comfortable working below library level: model code, custom losses, parallelism configuration.
• Experience with distributed training (FSDP, DeepSpeed or Megatron‑LM; data, tensor, pipeline and expert parallelism) on multi-node clusters with Slurm/K8s.
• Hands‑on post-training experience: SFT, DPO, and at least one RL method (PPO, GRPO or similar).
• Solid mathematical foundations in linear algebra, optimisation and probability, and an understanding of training dynamics.
• Good understanding of GPU fundamentals: memory hierarchy, mixed precision, profiling, throughput.
• Rigorous experimental practice: ablation design, reproducibility, experiment tracking, noise‑aware evaluation.
• Strong debugging and communication skills. Fluent English.

Will be a plus

We don’t expect anyone to have all of these. Our research covers many areas and nobody on the team is an expert in all of them. If you have deep experience in some and none in others, we still want to hear from you.

• Experience with the NVIDIA stack (Megatron‑LM, Megatron Bridge, NeMo RL, NeMo Gym, NeMo Curator) or similar frameworks (verl, TRL, OpenRLHF).
• RL infrastructure: rollout engines (vLLM, SGLang), asynchronous RL, reward model and verifier design, agentic environments.
• Model compression: post‑training quantization and QAT (GPTQ, AWQ, FP8/FP4), pruning, distillation, low‑rank or tensor‑network methods.
• Post‑transformer architectures: SSMs/Mamba, linear attention, hybrid models, diffusion LMs.
• Physical AI: vision‑language‑action models, world models (e.g. JEPA‑style or video world models), robotics simulation.
• Multimodal model training (vision‑language, audio).
• Inference on alternative hardware: ExecuTorch, llama.cpp, ONNX Runtime, OpenVINO, vendor NPU SDKs, custom kernels (Triton, CUDA).
• Large‑scale data processing and synthetic data generation.
• Evaluation tooling (lm‑evaluation‑harness, lighteval, NeMo Evaluator) and familiarity with Artificial Analysis.
• Publications at top venues (NeurIPS, ICML, ICLR, ACL, etc.), notable open-source contributions, or models released on Hugging Face.
• Spanish.