SC
AI/LLM Engineer (Model Fine-Tuning & Conversational AI)
Salve.Inno Consulting
Contract type
Freelance
Work mode
Remote · Europe
Experience
3 years
Job description
Hands-on engineering role at an international technology company developing AI-powered conversational products, focused on training, fine-tuning and improving self-hosted open-weight LLMs. Covers the full model improvement lifecycle from data preparation and experiments to production deployment.
At a glance
- Employment
- Contract · Independent contractor
- Work mode
- Remote · Europe
What you'll do
- Train, fine-tune and optimise existing open-weight language models to improve conversational performance
- Develop and maintain model post-training workflows using supervised fine-tuning and parameter-efficient training techniques
- Prepare, clean, structure and improve training datasets to enhance response accuracy, naturalness and consistency
- Design and implement evaluation frameworks to measure model quality and identify opportunities for improvement
- Improve conversational memory, contextual understanding and personality consistency across extended interactions
- Develop and optimise multilingual conversational capabilities
- Enhance interactions involving multiple AI characters and complex dialogue scenarios
- Investigate model failures, inconsistent responses and behavioural issues
- Optimise training data, prompting approaches and response filtering mechanisms based on evaluation findings
- Conduct GPU-based model training experiments and troubleshoot performance, memory and training stability issues
- Apply techniques such as SFT, LoRA and QLoRA, with exposure to preference optimisation approaches such as DPO or GRPO
- Analyse experimental results and identify measurable improvements in model behaviour and conversational quality
- Deploy improved models into production environments and monitor their performance
- Continuously refine deployed models using evaluation data, user interaction patterns and performance insights
- Collaborate with AI engineers, product teams, web developers and DevOps specialists on model integration and deployment
- Research and evaluate relevant developments in open-weight language models and post-training methodologies
What we're looking for
- Proven hands-on experience personally fine-tuning open-weight LLMs and deploying improved models into production
- Strong professional programming skills in Python
- Practical experience using PyTorch and Hugging Face or comparable machine learning frameworks
- Hands-on experience with supervised fine-tuning (SFT) and parameter-efficient fine-tuning techniques such as LoRA and QLoRA
- Familiarity with preference optimisation techniques, particularly DPO or GRPO
- Experience independently preparing training datasets and building model evaluation frameworks
- Practical experience conducting GPU-based training experiments, including debugging training issues
- Strong understanding of model behaviour, conversational quality and post-training optimisation
- Ability to investigate model weaknesses, implement improvements and demonstrate measurable results
- Experience taking model improvements from experimentation through to production deployment
- Previous experience working with consumer-facing AI applications or products
- Ability to work with complex conversational datasets and assess model outputs systematically
- Strong analytical thinking, technical problem-solving skills and attention to detail
- Ability to collaborate effectively with engineering, product and infrastructure teams
- Comfortable taking ownership of model training, evaluation and deployment tasks
- Ability to work independently in a fast-paced, international remote environment
- Based in Europe and willing to work under a B2B contractor agreement
Nice to have
- Experience fine-tuning models for conversational AI, AI companionship, roleplay or interactive entertainment
- Experience improving long-context dialogue, memory retention or personality consistency
- Practical experience with multilingual LLM training and evaluation
- Experience implementing preference-based optimisation using DPO or GRPO
- Familiarity with production model monitoring and iterative improvement workflows
- Experience optimising GPU training efficiency and troubleshooting model training infrastructure
- Experience with consumer generative AI products serving large user bases
Benefits
- Fully remote working arrangement within Europe
- Long-term collaboration under a B2B contractor agreement
- 20 working days of paid leave
- $100 monthly health insurance allowance
- Equipment budget to support your remote working setup
- Access to AI development tools
- Learning and professional development support
- Annual team gathering
- Opportunity to work directly with self-hosted LLMs powering consumer-facing conversational AI products
- Hands-on ownership of model training, evaluation and production improvements
- International working environment within a rapidly growing AI technology company
- Opportunity to contribute to innovative developments in conversational AI and LLM post-training