SoftQuantus is seeking a PhD-level researcher to join its internal research and development team working on large language models, deep learning systems and scientific artificial intelligence.
The successful candidate will contribute to the design, training, adaptation and optimization of advanced language models for enterprise and scientific applications. The role combines applied research, experimental development and production-oriented engineering.
Research areas include:
• distributed training of large language models on GPU and HPC environments;
• continued pre-training, supervised fine-tuning and instruction tuning;
• parameter-efficient adaptation methods such as LoRA and QLoRA;
• model quantization, pruning and inference optimization;
• mixture-of-experts and sparse model architectures;
• retrieval-augmented generation;
• scientific and technical language models;
• multimodal and tool-using models;
• agentic AI and multi-agent orchestration;
• evaluation of accuracy, reliability, hallucination, latency and computational cost;
• reproducibility and provenance of AI training experiments;
• private and enterprise-controlled LLM deployment.
RESPONSIBILITIES
The researcher will:
• design and execute LLM training and adaptation experiments;
• prepare, validate and document scientific or enterprise datasets;
• implement distributed training pipelines;
• optimize memory use, throughput and inference performance;
• evaluate different model architectures and training strategies;
• develop reproducible benchmark protocols;
• analyse model behaviour, limitations and failure modes;
• integrate trained models with SoftQuantus AI systems and agent architectures;
• contribute to technical reports, research publications and patentable developments;
• maintain source code, configurations and experiment evidence in company-approved repositories;
• collaborate with quantum computing, HPC and software engineering researchers.
REQUIRED PROFILE
Candidates should have:
• a PhD, submitted PhD thesis, advanced doctoral research experience or equivalent research-level expertise in artificial intelligence, machine learning, computer science, applied mathematics or a related field;
• strong experience with Python and PyTorch;
• practical knowledge of transformer architectures and LLM training;
• experience with GPU computing and distributed training;
• knowledge of model evaluation and experimental methodology;
• ability to document research clearly and reproducibly;
• professional working proficiency in English.
Experience with DeepSpeed, FSDP, Megatron-LM, Hugging Face, Ray, Kubernetes, Slurm, CUDA or European HPC systems is advantageous.
POSITION STRUCTURE
This is an internal industrial research position within SoftQuantus.
It is not currently an accredited university PhD studentship and does not independently award a doctoral degree. Candidates interested in pursuing formal doctoral registration may be supported in exploring future collaboration with an academic institution, subject to the establishment of a university partnership.
The initial contractual and compensation structure will depend on the candidate’s experience, location, availability and the funding status of the associated research projects. Possible structures may include research collaboration, project-based contracting, equity-linked participation or salaried employment following project funding.
LOCATION
Remote within Europe, with possible collaboration involving SoftQuantus activities in France and Estonia.
APPLICATION
Applicants should send the following information to partner@softquantus.com:
• curriculum vitae;
• short research statement;
• links to publications, repositories or relevant technical work;
• description of previous experience training or optimizing language models;
• expected availability;
• country of residence;
• preferred contractual arrangement.
Email subject:
Application — PhD-Level LLM Researcher
