LLM Agent Developer using LangChain Agentic Workflows

LLM Agent Developer

Key Responsibilities

1. Design & Build LLM Agents

• Design

  single-agent and multi-agent

  architectures using

  LangChain

  .
• Implement agents that can:
  
• Call internal/external APIs (CRMs, ERPs, ticketing tools, etc.).
  
• Query databases / vector stores and combine results.
  
• Execute multi-step workflows with planning, tool selection, and error recovery.
  

2. LangChain-based Orchestration

• Use

  LangChain core, LCEL, tools, agents, retrievers, and chains

  to build reusable components.
• Implement:
  
• Tools (custom tools for APIs, DBs, file systems, etc.).
  
• Memory (conversation memory, summary memory, entity memory).
  
• Callbacks / tracing for observability.
  

3. RAG & Knowledge Access

• Integrate

  RAG pipelines

  inside agents using:
• Vector DBs (pgvector, Chroma, FAISS, Milvus, Weaviate).
  
• Chunking, embeddings, and retrievers.
  
• Design prompts so agents correctly

  use tools vs direct generation

  .

4. Backend Integration

• Build

  Python-based backends (FastAPI / Flask)

  to expose agent workflows as APIs.
• Integrate with

  web apps, portals, chat UIs, CRMs, internal dashboards, or workflow tools

  .
• Handle authentication, rate limiting, logging, and error handling.
  

5. Evaluation, Monitoring & Optimization

• Track and improve

  latency, cost, and accuracy

  of agent workflows.
• Use tracing / analytics (e.g.,

  LangSmith or similar tools

  ) for debugging.
• Run A/B tests on prompts, tools, and routing strategies.
  

6. Collaboration & Documentation

• Work with product/ops teams to convert business flows into

  agent workflows

  .
• Document:
  
• Agent responsibilities
  
• Tool contracts (input/output schemas)
  
• Failure handling strategies
  
• Follow best practices for

  versioning, testing, and deployment

  .

Required Technical Skills

Core LLM & LangChain

• Strong hands-on experience with

  LangChain

  (must-have).
• Building:
  
• Tool-using agents (e.g., AgentExecutor, Tool, ChatOpenAI/llms).
  
• Custom tools: REST APIs, DB queries, file operations.
  
• Chains using LCEL / Runnable interfaces.
  
• Good understanding of

  prompt engineering, function calling / tool calling

  , and structured outputs (Pydantic / JSON).

Backend & Programming

• Strong in

  Python

  (async, typing, structuring larger projects).
• Experience with

  FastAPI

  or Flask for serving agent workflows as APIs.
• Working knowledge of

  Git, CI/CD

  , and basic testing practices (pytest).

RAG, Data & Integrations

• Experience with:
  
• Vector databases: pgvector / Chroma / FAISS / Milvus / Weaviate.
  
• Embeddings & chunking for document Q&A.
  
• Integrating 3rd party APIs (CRM, ERP, ticketing, payment, etc.).
  

Infrastructure & Ops (Basics)

• Comfort with

  Docker

  , virtual environments, and deploying Python services.
• Working on Linux-based environments.
  
• Understanding of environment management, secrets, and config separation.
  

Experience Required

• Total Experience:

  26 years in Software / Data / ML Engineering.

• Relevant LLM / LangChain Experience:

• Minimum

  1+ year

  building LLM-based applications and/or agentic workflows.
• At least

  1–2 real projects

  using LangChain in POC or production.

Preferred / Good-to-Have Skills

• Experience with:
  

• LangGraph

  , AutoGen, or other multi-agent frameworks.
• Tools like

  OpenAI, Anthropic, Azure OpenAI, local LLMs

  (Llama, Mistral, etc.).
• LLM observability tools (e.g.,

  LangSmith, Weights & Biases, Ragas, DeepEval

  ).
• UI frameworks for chat interfaces (React/Next.js, Streamlit, Gradio).
  
• Orchestration tools like

  Dify, Flowise, OpenWebUI

  , etc.
• Basic knowledge of:
  
• RDBMS (PostgreSQL, MySQL) and data modeling.
  
• Caching, rate limiting, and production reliability concepts.