IN_ Manager_ Agentic AI

Technical Leadership & Delivery

•            Lead a team of GenAI developers and backend engineers to deliver enterprise GenAI applications and accelerators.

•            Translate business requirements into technical designs, work breakdown, implementation plans, and engineering tasks.

•            Review architecture, code, prompt design, RAG pipeline design, data model, API design, and integration approaches.

•            Ensure quality, performance, security, maintainability, observability, cost efficiency, and production readiness.

Advanced RAG, Semantic Search & Knowledge Architecture

•            Design and optimize RAG pipelines including ingestion, chunking, metadata extraction, embeddings, vector indexing, hybrid retrieval, re-ranking, citation generation, and answer synthesis.

•            Implement semantic search across structured, semi-structured, and unstructured knowledge sources.

•            Integrate with enterprise knowledge repositories such as SharePoint, Confluence, S3, Google Drive, databases, ITSM tools, and document management systems.

•            Define knowledge freshness, access-controlled retrieval, lineage, deduplication, feedback loops, and retrieval quality improvement mechanisms.

Agentic AI & MCP-based Tool Integration

•            Design and implement agentic workflows using LangGraph, LangChain, CrewAI, AutoGen, LlamaIndex, or equivalent.

•            Implement tool calling, memory, workflow state, human-in-the-loop approval, API actions, and safe execution boundaries.

•            Build or guide MCP server implementations for secure agent-to-tool and LLM-to-enterprise-system integration.

•            Evaluate A2A communication patterns for multi-agent collaboration, delegation, and hand-offs where relevant.

Backend, API, Data & Event-Driven Engineering

•            Build scalable services using Python, FastAPI, Node.js, Java, or similar technologies.

•            Design REST APIs and microservices for LLM orchestration, semantic search, ingestion, agent execution, feedback, and evaluation.

•            Use SQL and NoSQL databases for application metadata, session history, audit logs, feedback, configurations, and workflow state.

•            Design Redis caching patterns for session state, memory, rate limiting, token optimization, retrieval caching, distributed locks, and workflow continuity.

•            Design asynchronous workflows using Kafka, RabbitMQ, Azure Service Bus, AWS SQS/SNS, Google Pub/Sub, or equivalent, including retry, DLQ, idempotency, ordering, and schema management.

Evaluation, Observability, Security & Governance

•            Implement LLM tracing, prompt/version tracking, token usage monitoring, latency tracking, cost monitoring, and feedback analytics.

•            Define evaluation datasets and metrics for faithfulness, groundedness, answer relevance, context relevance, hallucination rate, toxicity, latency, and cost.

•            Use LangSmith, Langfuse, RAGAS, DeepEval, TruLens, Promptfoo, OpenTelemetry, Grafana, Prometheus, ELK, Datadog, or