Job Description

Job Description & Summary

The opportunity

Build, test and integrate production-grade AI agents and services that execute business tasks reliably within enterprise workflows.

What you will be doing

· Implement agents, prompts, tools, retrieval pipelines and orchestration logic.

· Integrate AI components with enterprise APIs, applications, databases and workflow services.

· Build automated tests and evaluation datasets for functional and non-functional behavior.

· Diagnose model, retrieval, tool-use and integration failures.

· Contribute to secure coding, documentation, peer review and release activities.

· Participate actively in agile ceremonies, demonstrations and backlog refinement.

What we need from you

· 3+ years in software, data or AI engineering.

· Strong Python or comparable programming skills, API development and version control.

· Practical experience with LLM applications, RAG, agents, embeddings and structured outputs.

· Ability to work iteratively with product, architecture, data and user-experience specialists.

Relevant AI technologies and tooling

· Hands-on experience building agents with at least one production-oriented framework such as LangChain and LangGraph, Microsoft Agent Framework or Semantic Kernel, OpenAI Agents SDK, AutoGen, CrewAI, or equivalent.

· Strong Python skills and practical experience with FastAPI or similar API frameworks, Pydantic or comparable schema validation, asynchronous programming, Git and automated testing.

· Practical experience implementing tool calling, structured outputs, agent state and memory, hand-offs, guardrails, retries, human-in-the-loop steps and deterministic workflow nodes.

· Experience implementing RAG pipelines using embeddings, vector or hybrid search, metadata filters, reranking and evaluation datasets.

· Familiarity with MCP, enterprise API integration, queues or events, containerization with Docker and deployment to Kubernetes or managed application platforms.

· Ability to instrument agent executions using tracing and evaluation tools such as LangSmith, MLflow, Langfuse, OpenTelemetry or platform-native equivalents.

Measures of success

· Working features delivered per iteration

· Evaluation results and defect rates

· Integration reliability

· Code review and documentation quality

· Contribution to reusable engineering assets

Key interfaces

· Other members of the AI Transformation & Agentic Systems Practice

· PwC sector, functional, cloud, cyber, risk, Responsible AI and change specialists

· Client business owners, product owners, technology teams and operational users

· Technology alliance and implementation partners where relevant

Contribution to the practice

· Support proposals, client workshops and market development appropriate to seniority.

· Contribute reusable methods, patterns, code, assets and lessons learned.

· Coach colleagues and participate in the capability’s continuous learning agenda.

· Uphold PwC quality, independence, confidentiality and risk-management requirements.

#LI-BS1 #LI-Hybrid