AI Engineer - LLM Applications & Prompt Engineering

We are looking for an AI Engineer to build and improve the AI experiences across our education platform, including personalised tutoring, test preparation, college guidance, research and writing support.

This is a hands-on engineering role with a substantial focus on prompt development and tuning. You will translate educational and product requirements into dependable AI behaviour, build the services and agent workflows behind it, and take changes from experimentation through to production. You will work closely with the CTO, product team, subject specialists and software engineers.

Responsibilities

Prompt engineering and tuning

• Own extensive prompt development, testing and continuous refinement across our AI features, improving accuracy, relevance, clarity, tone and consistency.
• Design system prompts, reusable templates, few-shot examples, tool instructions and structured output schemas. Refine how agents ask questions, explain concepts, provide feedback and guide students through tasks.
• Manage the context supplied to models, including conversation history, learner profiles, activity, current workspace and relevant documents or school information.
• Analyse real conversations and feedback with product and education specialists. Turn ambiguous requirements and recurring failure cases into measurable improvements.

AI agents and application engineering

• Build and maintain Python services, conversational agents and workflows that use tools, retrieve information, generate reports and take authorised actions through platform APIs.
• Integrate LLM providers and supporting services; evaluate model choices and configure routing, token budgets, streaming and fallbacks.
• Develop retrieval and grounding approaches that give AI responses the relevant information and preserve source attribution where needed.
• Connect AI workflows to learner state, task management and platform events. Implement explicit business rules, permissions and reliable execution for actions that change user data.

Evaluation and production quality

• Build evaluation datasets and automated prompt regression tests covering answer quality, educational usefulness, factual accuracy, tool use and multi-turn behaviour.
• Version prompts and model configurations, compare changes against a baseline, and document results so releases can be reviewed and reproduced.
• Investigate production failures and improve observability, error handling, retries and recovery. Monitor latency, token usage, cost and task completion.
• Test prompt injection, inappropriate requests and other adversarial inputs. Implement guardrails and appropriate uncertainty handling for a platform used by students.

Requirements

• Strong Python engineering skills and experience building, testing and maintaining production services and APIs.
• Practical experience shipping LLM applications, with examples of substantial prompt tuning and evidence of how you measured improvements.
• Experience with tool calling, structured outputs, conversational state, context management and agent orchestration.
• An understanding of retrieval, embeddings and grounding, and when these approaches help solve a product problem.
• A systematic approach to evaluation and debugging, including the ability to distinguish prompt failures from data, model, integration or application issues.
• Strong written English and the ability to turn complex product and educational requirements into precise instructions and clear technical decisions.
• Ownership of delivery, sound judgement about quality, latency and cost, and effective collaboration with technical and non-technical colleagues.

Nice to have

• Voice, image or other multimodal AI experiences; model fine-tuning or small-model classification and routing.
• TypeScript, event-driven systems, cloud deployment, Docker and AI evaluation or observability tooling.
• Education technology, AI tutoring, assessment, feedback generation or personalised learning.

We offer

• Fully remote. No office, no commute. We work asynchronously where we can and synchronously where we must. You'll have meaningful overlap with the both founder and CTO (London) and the rest of the team across London and the US.
• Professional growth opportunities and career prospects
• 25 days plus public holidays and standard employer contribution depending on location.
• Any subscription you want - Claude Max, ChatGPT Pro, whatever your stack needs.

About Dakota

Dakota.Education is building the world's most exciting new education brand. We make AI-native products that teach K–12 students to read, write and think, and help US high schoolers prep for the SAT and write college essays that sound like them. We've partnered with Penguin Random House, HarperCollins and MIT Technology Review. Our advisors come from College Board, ACT, Stanford and Dartmouth. We're a Google Accelerator alumna.

We have a clear, opinionated view: the answer to the reading recession and the college admissions arms race isn't another app — it's an education company, built for this moment. We believe AI, built right, can help any student reach their outcomes and their dreams without offloading the thinking — uplevelling them rather than replacing them. And we believe reading sits at the heart of everything, including college readiness.

Claude and Gemini are wired into how we work. Every team member uses it daily. We expect you to add to that infrastructure, not catch up to it.

Think Claude, but purpose-built to deliver outcomes on reading, test prep and college admissions.

Application

This role requires you to be eligible to work in one of these locations: Poland, Estonia, the UK, or the USA.

Apply on our careers website: https://careers.dakota.education/jobs/c5b14d33-efef-4240-81f4-965676d8e8b4/apply