AI Engineer/ML Engineer - Senior Developers - AI Training - Oxford, UK

Role overview Contribute as a senior AI/ML engineer in an expert network that supplies research teams with high-calibre technical judgement for training and evaluating large language models. The role is remote and task-based, attached to the Oxford, UK region, and focuses on improving model accuracy, safety, and reasoning quality.

Responsibilities

- Assess AI-generated technical content on model architectures, loss functions, and backpropagation for correctness. - Review ML codebases and notebooks, validating training loops, data pipelines, and evaluation logic. - Deliver rigorous human feedback that informs RLHF and alignment efforts. - Interrogate chain-of-thought reasoning, surfacing hallucinations, biased outputs, and breakdowns in logic. - Perform comparative benchmarking of model outputs against defined technical taxonomies and metrics.

Requirements

- BS, MS, or PhD in Computer Science, Artificial Intelligence, Robotics, or a closely related quantitative field centred on machine learning. - Track record of building, deploying, or fine-tuning ML models in production. - Professional understanding of neural network architectures (Transformers, CNNs, RNNs) and modern optimization methods. - Experience with LLM-specific workflows including prompt engineering, RLHF, or RAG. - Ability to audit model logic, detect data contamination, and evaluate the mathematics behind ML algorithms. - Careful, sceptical eye for catching subtle hallucinations and reasoning failures.

Nice to have

- Expert-level command of PyTorch or TensorFlow/Keras, advanced Python (NumPy, Pandas, Scikit-learn), and Hugging Face Transformers. - Familiarity with AWS SageMaker, Google Vertex AI, or tools such as Weights & Biases, LangChain, Pinecone, Milvus, or Weaviate.

Benefits and work setup

- Fully remote, flexible contributor arrangement with paid hourly tasks, often paying up to $80 per hour for this profile. - Typical tasks require around an hour of focused, uninterrupted work, though many are shorter.