Senior AI Engineer

About the Role The Full Stack AI Consultant (Developer) is a hands-on engineer at the core of Accenture's agentic AI delivery capability. This role is for those who build things — full stack applications, agentic workflows, knowledge pipelines, and tool integrations — and who want to do that work at the frontier of enterprise AI. Consultants work within delivery teams, turning business requirements into production-grade agentic AI solutions under the guidance of technical leads and managers. The expectation is active, daily engineering: writing code, building agents, implementing MCP servers, designing knowledge pipelines, and maintaining the DevOps and AgentOps practices that keep production systems running. Consultants on this team are also expected to invest continuously in learning — agentic AI is evolving rapidly, and staying current is a professional responsibility, not optional. Position Responsibilities Full Stack Application Development • Design and build full stack agentic AI applications — Python backends, REST and event-driven APIs, and React or equivalent frontends — to production engineering standards. • Implement agentic application UX: streaming responses, intermediate output display, reasoning transparency, and error and escalation interfaces for end users. • Translate business requirements into technical specifications; work with managers and clients to clarify scope, surface ambiguities, and deliver against agreed outcomes. Agentic AI Development • Build and configure agents using established orchestration frameworks (LangGraph, AutoGen, or equivalent): harness setup, persona and instruction loading, tool binding, memory configuration, and lifecycle management. • Implement reasoning patterns (ReAct, Chain-of-Thought, Plan-and-Execute) appropriate to each agent use case; design and version prompt architecture including system prompts, few-shot examples, and structured output schemas. • Build multi-agent workflows— defining agent roles, A2A handoff contracts, shared state schemas, and escalation paths — under architectural guidance from technical leads. MCP, Tools, Skills, and Workflows • Design, build, and maintain MCP servers connecting agents to enterprise systems, APIs, databases, and SaaS platforms — with robust schema design, error handling, idempotency, and retry logic. • Translate business processes into agent-executable skills, structured instructions, and reusable workflows — bridging the gap between business requirements and agent implementation. • Implement context engineering pipelines, memory architectures (episodic, working, long-term), andLLM gateway configuration to support reliable, cost-efficient agent operation. Knowledge Layer Implementation • Build RAG pipelines: document ingestion, chunking, embedding, vector store indexing, hybrid retrieval, re-ranking, and quality evaluation. • Implement Text-to-SQL capabilities — schema grounding, query generation, validation, and safe execution against enterprise databases. • Integrate Elasticsearch as a retrieval backend; build knowledge graph components and ontology-driven query layers where required by the use case. DevOps, AgentOps, and Quality • Maintain CI/CD pipelines for agent code, prompt changes, and infrastructure — including automate devaluation gates and deployment strategies across environments. • Instrument agentic systems with production observability: distributed tracing, token cost tracking, latency profiling, failure logging, and drift detection. • Build agent testing suites: unit tests with mocked tools, multi-agent integration tests, and simulation environments; implement guardrails, PII redaction, and audit trail logging. • Manage versioned agent and asset registries — agents, tools, skills, prompts, and workflows — with controlled promotion across development, staging, and production. Continuous Learning and Innovation • Maintain active, current knowledge of agentic AI frameworks, tooling, and research — testing new approaches and bringing relevant innovations into the team's engineering practice. • Contribute to internal knowledge sharing: documenting patterns, building reusable accelerators, and supporting capability development within the team.