Agentic AI Developer

Agile Bridge is looking for an Agentic AI Engineer who can connect foundation models, enterprise knowledge and software tools into secure, reliable AI solutions. You will design and build flows, agents, skills and evaluations that do useful work within real business processes, not isolated chat demonstrations.

What to expect in this role

A project may require you to map a business workflow, decide where AI should and should not act, create skills that connect an agent to enterprise APIs, build a grounded retrieval layer, manage context or memory, and define the evaluation set that proves the solution works. You will design explicit permission boundaries, human approval points and escalation paths so that increasing autonomy does not create uncontrolled risk.

This is a software- and solution-engineering role built around available models. You will collaborate with AI Engineers when custom model work or inference optimisation is needed, Data Scientists when statistical validity or analytical insight is central, and Software and Platform Engineers on enterprise integration and production operations.

What you will do

• Translate business processes into clear flows, agent responsibilities, skills, decision points and measurable success criteria.
• Design agentic architectures that integrate foundation models, enterprise applications, APIs, data sources and human controls.
• Develop and test flows, agents, skills, services and APIs using strong software-engineering practices.
• Build retrieval, context, memory and state-management patterns that provide relevant, grounded information.
• Create evaluation datasets and automated checks for task completion, grounding, tool use, safety, latency and cost.
• Deploy and monitor AI services, using traces and operational evidence to improve prompts, retrieval, context and orchestration.
• Implement guardrails, access controls, privacy protections and secure tool-use patterns.
• Communicate behaviour, trade-offs, limitations and operational requirements to customers and engineering teams.
• Research emerging models and frameworks, adopting them only where evaluation demonstrates meaningful value.

What you need

• A bachelor's degree in Computer Science, Software Engineering, Information Technology, Data Engineering, Data Science, Artificial Intelligence or a related field.
• Typically 3-5 years' relevant experience in software or AI engineering, including production software delivery.
• Strong Python capability and evidence of clean, testable and maintainable engineering practices.
• Practical experience integrating foundation-model capabilities into applications, services or workflows.
• Experience with APIs, enterprise data integration and cloud-based delivery practices.
• The judgement to balance usefulness and autonomy with reliability, security, privacy, latency and cost.

Core technical skills

We do not expect experience with every named framework. We do expect strong capability across the underlying engineering concepts and the ability to evaluate tools objectively.

• Software engineering: Python, modular design, automated testing, debugging, Git, code review, CI/CD and maintainable service development. TypeScript or C# is useful in integration-heavy environments.
• APIs and integration: REST APIs, authentication, microservices or distributed-system concepts, enterprise data integration and safe tool contracts.
• Foundation-model applications: Model selection, prompt engineering, structured outputs, tool or function calling, context-window management and failure handling.
• Agentic design: Flows, agent responsibilities, skills, state, planning patterns, tool permissions, human-in-the-loop controls and orchestration.
• Retrieval and memory: RAG, embeddings, chunking, search and reranking, citations or grounding, vector stores, and context or memory patterns with appropriate retention controls.
• Evaluation: Test cases, evaluation datasets, rubrics, automated scoring, trace review, regression testing and measurement of quality, safety, latency and cost.
• Cloud and delivery: Docker, CI/CD, configuration and secrets management, cloud AI services, observability, logging and production support.
• Security and governance: Prompt-injection and data-leakage controls, privacy and PII protection, least privilege, auditability, guardrails and safe escalation or failure behaviour.

Useful, but not essential

• An Honours or Master's degree in Artificial Intelligence, Computer Science, Software Engineering or a related field.
• Relevant Microsoft Azure, AWS or Google Cloud AI certification.
• Experience with MCP and an orchestration framework such as Semantic Kernel, LangGraph, LangChain, CrewAI or an equivalent.
• Experience with n8n, Make or comparable automation tooling.
• Exposure to PostgreSQL, vector databases, graph databases, Kubernetes, vLLM, Ollama or private-model deployment patterns.
• Experience implementing enterprise identity, authorisation and approval workflows for AI tools.

How you work

• You understand the business process before deciding where an agent belongs.
• You design autonomy deliberately and do not confuse impressive demos with safe production capability.
• You test behaviour systematically and use evidence to improve the solution.
• You communicate uncertainty, limitations and operational trade-offs honestly.
• You collaborate across business, software, data, AI and platform disciplines.

The opportunity

You will join an innovation-focused engineering environment where agentic AI must deliver measurable value under real enterprise constraints. The role offers exposure to orchestration, enterprise retrieval, tool-enabled agents, evaluation, responsible autonomy and production LLMOps while working with experienced engineering teams.

How success will be measured

• Flows, agents and skills address a clearly defined business need and complete intended tasks reliably.
• Outputs and actions are grounded, traceable and appropriately controlled.
• Evaluation evidence demonstrates acceptable quality, safety, latency and cost before release.
• Production behaviour remains observable, with failures, regressions and emerging risks addressed early.
• Solutions remain maintainable and integrate effectively with enterprise systems and operating processes.