Location: [On-site] Employment Type: Full-time Experience: 1-3 years
About Intugle
Intugle is a GenAI-powered platform that helps enterprise organizations to transform their legacy business applications with AI embedded intelligence.
About the Role
We are looking for an AI Engineer with hands-on experience building and deploying production-grade AI agents. You will design, develop, and operate multi-agent systems using the LangChain ecosystem (LangChain, LangGraph, LangSmith), integrating LLMs, tools, memory, and external APIs to automate complex workflows.
This is a high-ownership role — you will work across the full stack from agent design through deployment and observability.
Key Responsibilities
- Design and implement single and multi-agent pipelines using LangChain and LangGraph, covering orchestration, state management, and tool use
- Build custom tools, retrievers, and memory modules to extend agent capabilities
- Integrate agents with external systems via REST APIs, MCP servers, and function-calling interfaces
- Implement RAG pipelines — from chunking strategy and embedding selection to vector store indexing and retrieval optimization
- Write and refine prompts; maintain prompt versioning and regression test suites
- Instrument agents with LangSmith (or equivalent) for tracing, evaluation, and latency profiling
- Deploy agents to cloud infrastructure (AWS, GCP, or Azure) using containerized services (Docker, Kubernetes)
- Collaborate with product and domain teams to translate business workflows into agentic solutions
- Participate in code reviews, architecture discussions, and documentation
Required Skills
Core Agent Engineering
- LangChain (chains, agents, tools, memory, callbacks)
- LangGraph (stateful multi-agent graphs, conditional edges, human-in-the-loop)
- Prompt engineering and structured output design
- Tool/function calling with OpenAI, Anthropic, or open-source models
LLM & ML Foundations
- Working knowledge of transformer-based LLMs and their API surface (context windows, temperature, sampling)
- Embedding models and similarity search
- Familiarity with fine-tuning concepts (LoRA, PEFT) — even if not hands-on
RAG & Knowledge Retrieval
- Vector databases (Pinecone, Weaviate, pgvector, Chroma, Qdrant)
- Chunking strategies, hybrid search, re-ranking
- Semantic layer design or ontology mapping (a plus)
Software Engineering
- Python (strong) — async patterns, typing, packaging
- REST API design and consumption
- SQL — ability to write and debug complex queries
- Git, CI/CD pipelines, environment management
Infrastructure & Deployment
- Docker and container orchestration basics
- Cloud deployment: AWS Lambda / ECS / EKS, or GCP Cloud Run, or Azure Container Apps
- Environment and secrets management
- Basic understanding of message queues (Kafka, SQS, Redis) for async agent pipelines
Observability & Evaluation
- LangSmith or equivalent tracing tools
- Evaluation frameworks for LLM outputs (LLM-as-judge, RAGAS, custom evals)
- Logging, alerting, and SLA monitoring for AI systems
Nice to Have
- Experience with Model Context Protocol (MCP) — building or consuming MCP servers
- Knowledge of Agent-to-Agent (A2A) communication patterns
- Graph databases (Neo4j) for semantic/schema retrieval
- AST-based code analysis or data lineage tooling
- Experience deploying on enterprise platforms (Azure ML, AWS SageMaker, Vertex AI)
- Contributions to open-source AI/ML projects
