Senior / Expert AI Engineer — Agentic AI, RAG, LLM Architecture & Python

## Project Overview We're building an AI-powered educational platform and looking for an experienced, hands-on AI engineer to help design and implement its core AI systems. The application and broader product development are being handled internally. We're specifically looking for someone who can take technical ownership of the **AI architecture and implementation**, help make the right engineering decisions, and build reliable, production-ready AI capabilities. This is not a basic chatbot or prompt-engineering project. The platform will involve conversational AI, retrieval-augmented generation, intelligent agents, personalised interactions, structured knowledge, and systems that maintain and use contextual information over time. We're looking for someone who has already built complex AI systems and understands what it takes to move beyond prototypes into reliable, scalable applications. ## What You'll Work On ### 1. AI Architecture & System Design - Evaluate existing technical approaches and recommend suitable AI architecture. - Design modular, scalable AI services that integrate with an existing application. - Make architectural decisions around LLM orchestration, retrieval, agent workflows, memory, and state management. - Evaluate frameworks and technologies based on practical requirements rather than trends. - Help identify potential bottlenecks, trade-offs, and implementation risks. ### 2. Advanced RAG & Knowledge Retrieval - Design and implement robust RAG pipelines. - Handle document processing, chunking, embeddings, metadata and indexing. - Implement hybrid retrieval, reranking, query transformation and context optimisation. - Work with structured and unstructured knowledge sources. - Explore graph-based retrieval and knowledge graphs where they provide genuine value. - Improve retrieval relevance, factual accuracy and grounding. ### 3. Agentic AI & Multi-Agent Workflows - Build intelligent agents with tool calling, planning and multi-step execution. - Implement agent orchestration using LangGraph, OpenAI Agents SDK, LlamaIndex or suitable custom approaches. - Design specialised agent workflows with shared context and state. - Implement conditional routing, iterative reasoning, validation and controlled execution loops. - Develop reliable tool integrations and structured outputs. - Handle agent failures, retries, execution limits and workflow observability. ### 4. Memory, Personalisation & State Management - Design systems for maintaining relevant context across interactions. - Implement short-term and long-term memory strategies. - Work with structured user state, interaction history and evolving knowledge. - Build workflows that use previous interactions and outcomes to influence future recommendations. - Evaluate suitable approaches for storing, retrieving and updating contextual information. ### 5. AI Evaluation, Quality & Reliability - Design evaluation frameworks for RAG and agent workflows. - Create datasets and automated tests to measure output quality. - Evaluate retrieval precision, relevance, groundedness and task completion. - Address hallucinations, inconsistent responses and agent failure cases. - Implement tracing, monitoring and debugging for complex AI workflows. - Improve latency, token usage, reliability and operational costs. ### 6. Implementation & Integration - Develop clean, maintainable AI backend services using Python. - Build APIs that expose AI capabilities to the existing application. - Integrate LLM providers such as OpenAI, Anthropic, Google Gemini or AWS Bedrock. - Work with relational databases, vector stores and graph databases where appropriate. - Collaborate on deployment, testing and integration with the application backend. - Document architectural decisions and implemented workflows. ## Required Experience We're primarily looking for someone with **strong, demonstrable experience building and shipping production AI applications**. - 5+ years of professional software engineering experience preferred. - Strong hands-on expertise in Python. - Proven experience architecting and implementing production RAG systems. - Experience with LLM-based agents, tool calling and multi-step orchestration. - Strong understanding of embeddings, vector search, hybrid retrieval and reranking. - Practical experience with LangGraph, LlamaIndex, OpenAI Agents SDK or similar frameworks. - Familiarity with state management, memory architectures and persistent context. - Experience with LLM evaluation, observability, testing and production debugging. - Solid backend engineering skills, preferably Python and FastAPI. - Experience working with PostgreSQL and vector databases. - Ability to independently evaluate trade-offs and implement reliable solutions. **We value real production engineering experience significantly more than familiarity with a long list of frameworks.** ## Nice to Have - Experience with knowledge graphs, Neo4j or GraphRAG. - Experience with adaptive systems, recommendation engines or personalised AI. - Familiarity with knowledge tracing, learner modelling or educational AI. - Experience with multimodal documents, OCR and document understanding. - Familiarity with MCP and external tool integrations. - Experience building enterprise-grade AI systems. - Experience with AWS or Azure AI infrastructure. - Strong understanding of LLM security and data privacy. Direct EdTech experience is helpful but not required. ## What We're Expecting We're looking for someone who can contribute at both the **architecture and implementation levels**. You should be comfortable: - Understanding a product requirement and proposing a technically sound approach. - Explaining why a particular architecture or framework makes sense. - Implementing the proposed solution yourself. - Reviewing and improving existing approaches where necessary. - Working independently while collaborating with the founder/development team. - Thinking about evaluation, maintainability and production reliability from the beginning. **We're not looking for a consultant who only provides architecture diagrams or recommendations. We need someone who can actually build and take ownership of the AI components.** ## Division of Responsibilities **Our side:** - Product direction and requirements - Application frontend and user experience - General application backend and business workflows - Overall product integration and coordination **Your side:** - Core AI architecture - AI-specific backend services and implementations - RAG, agent orchestration, context and memory systems - AI evaluation, optimisation and reliability - Technical guidance on AI-related decisions You'll collaborate with us to ensure everything integrates cleanly into the existing application. ## Initial Engagement & Deliverables We'd prefer to start with a focused, paid technical engagement before considering a longer collaboration. Initial work would involve: 1. Understanding the current application and AI requirements. 2. Reviewing the proposed AI approach and identifying improvements. 3. Defining a practical architecture and phased implementation plan. 4. Implementing an agreed core AI workflow or component. 5. Demonstrating its functionality, quality and integration readiness. Based on the collaboration, we'd like to continue developing the remaining AI capabilities together. ## How to Apply Please share: 1. **Relevant experience:** Briefly describe the most complex AI or LLM system you've personally built. 2. **Production projects:** Share one or two real examples involving RAG, agents, memory or intelligent workflows. GitHub repositories, demos or case studies are welcome. 3. **Your contribution:** What parts did you personally architect and implement? 4. **Technical challenges:** Describe one difficult production AI issue you encountered and how you solved it. 5. **Architecture experience:** Have you designed an AI system from scratch, or significantly improved an existing architecture? 6. **Availability:** Your weekly availability, hourly rate and interest in ongoing collaboration. Please focus on your actual engineering contributions rather than providing a generic list of AI tools and frameworks. **We're looking for a technically strong, hands-on engineer who enjoys solving challenging AI problems and helping build a meaningful product over the long term.**