Remote AI Engineer – EdTech, Equity

Oliver Bernard; Greater London; Hybrid; GBP 75,000 - 85,000; Full time

Job summary

Oliver Bernard is seeking an AI Engineer for a remote role within the UK, at an early-stage EdTech startup focused on AI-powered assessment. You will build and refine AI-assisted marking, feedback and assessment workflows from requirements to production.

You will work with production Python, LLMs, and related AI systems, evaluating non-deterministic outputs and balancing quality, latency and cost. Some in-person meetups in London may occur.

Qualifications

• Experience building and deploying LLM/ML-based systems in production.
• Experience designing evaluations or regression tests for non-deterministic systems.
• Ability to investigate failures systematically and trade off quality, latency and cost.

Responsibilities

• Deliver AI-assisted assessment workflows from requirements through production and ongoing improvement.
• Translate marking guidance into rubrics, schemas, prompts and LLM programs.
• Build and maintain production Python code for assessment workflows and shared AI components.
• Design evaluation methods using representative datasets, regression tests, and statistical analysis.
• Investigate failed or contested assessments and improve workflows accordingly.
• Build human-review and feedback-driven optimisation workflows.
• Extend multimodal assessment capabilities across documents, images and video.

Skills

LLM systems

Regression testing

Failure analysis

AI/ML engineering

Python production

Education

Postgraduate degree in AI/ML

Job description

Oliver Bernard is seeking an AI Engineer for a remote role within the UK, at an early-stage EdTech startup focused on AI-powered assessment. You will build and refine AI-assisted marking, feedback and assessment workflows from requirements to production.

You will work with production Python, LLMs, and related AI systems, evaluating non-deterministic outputs and balancing quality, latency and cost. Some in-person meetups in London may occur.