> Markdown version of [/jobs/ext/3838334-senior-ai-engineer](https://www.wearedevelopers.com/jobs/ext/3838334-senior-ai-engineer). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md).

---

# Senior AI Engineer

- **Company:** Insight
- **Location:** London, UK
- **Experience:** Expert
- **Salary:** £56,191.0
- **Contract:** Permanent contract
- **Skills:** .NET Framework, Artificial Intelligence, Automation of Tests, Microsoft Azure, C Sharp (Programming Language), Computer Programming, Continuous Integration, Dependency Injection, Desktop Publishing, Fraud Prevention and Detection, Graph Database, Python (Programming Language), Neo4j, OpenAI, Software Engineering, SPARQL, Microsoft Agent Framework, Ontology Design, Large Language Models, Multi-Agent Systems, Generative AI, Agentic-AI, Api Design, Software Version Control, Data Pipelines
- **Published:** October 20, 2026
- **Apply:** https://www.adzuna.co.uk/jobs/details/5934455275

## About the Role

12+ years in software engineering; recent experience with generative AI applications and AI agents on cloud.

 - 1+ year hands-on with Microsoft Agent Framework (agent orchestration, tool calling, workflows).
 - Advanced skills in C# and .NET (async programming, dependency injection, API design).
 - Experience with Azure OpenAI/Microsoft Foundry and Azure cloud services.
 - Proficiency in knowledge graphs (Neo4j/Stardog, Cypher, SPARQL, ontology design).
 - Experience with event/time-series data and correlating events across sources.
 - Understanding of fraud and anomaly detection methodologies.
 - Experience designing evaluation/testing for LLM and agent outputs.
 - Ability to work with business/data/tech stakeholders and in regulated environments.

Desirable Skills

 - Experience with graph data science (community detection, link prediction, graph embeddings).
 - Python skills for data science/model experimentation.
 - Familiarity with identity/document verification signals.
 - Knowledge of model risk management, responsible AI, and UK/EU AI regulations.
 - Relevant certifications (Microsoft Azure AI Engineer Associate, Neo4j Certified Professional).

## Description

Design, build, and deploy fraud detection agents in C# using Microsoft Agent Framework.

 - Implement multi-agent workflows, tool calling, and human-in-the-loop approval processes.
 - Develop event-sequence detection to identify fraud patterns in account activities.
 - Build deterministic rules and risk-scoring logic for MVP, including case prioritization.
 - Transition detection logic from deterministic rules to generative AI methods.
 - Integrate anomaly reasoning, alert summarization, and agent-led investigation.
 - Ensure every alert includes explainability for fraud analysts (events, scores, evidence).

AI Training Agents

 - Build AI training agents to generate and curate labeled fraud scenarios and synthetic datasets.
 - Develop feedback loops using analyst decisions to tune rules, prompts, and models.
 - Create evaluation frameworks to measure detection rates, false positives, and reasoning quality.

Knowledge Graphs

 - Design, build, and maintain knowledge graphs in Neo4j/Stardog linking accounts, devices, transactions, etc.
 - Model graph ontologies and create queries (Cypher, SPARQL) to detect shared identifiers and fraud rings.
 - Integrate knowledge graph context into agent reasoning via retrieval and graph-based features.

Engineering & Delivery

 - Collaborate with Solution Architect and Business Analyst to translate fraud scenarios and data requirements into solutions.
 - Build data pipelines to ingest authentication, recovery, account, and transaction events.
 - Apply software engineering best practices: version control, CI/CD, automated testing, monitoring, cost & latency management.
 - Ensure compliance with security, privacy, model risk, and audit requirements for financial crime systems.
 - Document design decisions and hand over solutions to support teams.