Research Engineer

About the role

We are looking for a Research Engineer to join Soniox in Ljubljana and help shape the future of real‑time, multilingual speech and language AI. You will work at the intersection of cutting‑edge research and high‑performance engineering, turning state‑of‑the‑art models into production‑grade systems used by developers and enterprises worldwide.

Key responsibilities

• Deploy and optimise transformer‑based models for real‑time inference across speech recognition, translation and language understanding.
• Collaborate with researchers, engineers and product leaders to design experiments, run evaluations and ship features that improve user experience and performance.
• Build and optimise training, fine‑tuning and evaluation pipelines to continuously improve model performance and scalability.
• Own the end‑to‑end system, from prototyping new architectures to monitoring deployed models in production.
• Stay up‑to‑date with the latest AI advances, experiment with novel techniques and help maintain Soniox’s lead in accuracy, speed and efficiency.

Required profile

• Strong background in deep learning, NLP or speech technologies, with experience in transformer models (LLMs, ASR, MT, etc.).
• Proficiency in PyTorch and familiarity with distributed training, fine‑tuning techniques such as LoRA, distillation or RLHF, and inference optimisation.
• Experience building real‑time or high‑throughput systems and a focus on latency, reliability and performance.
• Ability to work in a fast‑paced environment, own loosely scoped problems and see them through from prototype to production.
• Master’s or PhD in Computer Science, AI or a related field, or equivalent practical experience.

Required skills

• Deep learning
• Natural Language Processing (NLP)
• Speech technologies / ASR
• Transformer models
• PyTorch
• Distributed training
• Fine‑tuning techniques (LoRA, distillation, RLHF)
• Inference optimisation
• Real‑time / high‑throughput system development