Senior LLM Engineer

About the role

Multiverse Computing, a well-funded deep-tech company founded in 2019 and described as the largest quantum software company in the EU, is looking for a Senior LLM Engineer to join its Research team in Barcelona. You will work across the full model lifecycle: data, distributed training, post-training, compression, evaluation and efficient inference. The company values strong fundamentals over narrow specialisation and publishes papers and releases models and code.

Key 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, QAT, tensor-network methods) and healing recipes for 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).
• Build data pipelines for curation, filtering, deduplication and large-scale synthetic data generation.
• Design ablations and evaluations on benchmark suites and internal evals.
• Port and optimise models for inference on accelerators and edge chips, profiling latency, memory and throughput.
• Write up experiments as internal reports, papers and open-source releases, review code, mentor teammates and follow the latest research.

Required profile

• Master's or PhD in CS, 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 mathematical foundations in linear algebra, optimisation and probability, and good understanding of GPU fundamentals.
• Rigorous experimental practice, strong debugging and communication skills, fluent English.

Required skills

• PyTorch, including model code, custom losses and parallelism configuration.
• Distributed training (FSDP, DeepSpeed or Megatron-LM) on multi-node clusters with Slurm/K8s.
• Hands-on post-training: SFT, DPO and at least one RL method (PPO, GRPO or similar).
• Preferred: NVIDIA stack (Megatron-LM, NeMo RL, NeMo Gym, NeMo Curator), vLLM, SGLang, GPTQ/AWQ/FP8, QAT, SSMs/Mamba, and evaluation tooling such as lm-evaluation-harness and lighteval.

What we offer

• Indefinite contract, variable performance bonus and signing bonus.
• Work visa sponsorship and relocation package (if applicable), private health insurance and eligibility for an educational budget.
• Hybrid opportunity, flexible working hours, career plan, language classes and discounted lunch options.