Lead Agentic AI Engineer Architect

Merkle; Gurugram District; On-site; INR 2,500,000 - 5,000,000; Full time

Job summary

Dentsu in Bengaluru (DGS India) is seeking a senior Python/LLM engineer to design and implement scalable AI solutions. You will work on agent orchestration, RAG, and modern AI tooling in a fast-moving team, delivering iterative improvements on complex distributed systems.

The role emphasizes hands-on development with LLMs, API design, and cloud platforms, with a permanent, full-time arrangement in a dynamic marketing/advertising tech environment.

Qualifications

• 4+ years of experience with strong Python skills.

Skills

Python

LLMs

Agent orchestration

API design

Microservices

Event-driven architecture

Azure

Vector databases

Embeddings

Distributed systems

Team collaboration

Tools

LangChain

LangGraph

AutoGen

CrewAI

Databricks

Genie

Job description

Job Description:

Required Qualifications

• 4+ years of experience with strong Python skills

• 2+ year hands-on experience building with LLMs (prompt engineering, RAG, function calling, agent frameworks)

• Experience with at least one agent orchestration framework (LangChain, LangGraph, AutoGen, CrewAI)

• Understanding of API design, microservices, and event-driven architectures

• Experience with cloud platforms (Azure preferred)

• Familiarity with vector databases, embeddings, and retrieval-augmented generation patterns

• Strong debugging and problem-solving skills in distributed systems

• Comfort working in a fast-moving team that ships iteratively

Preferred Qualifications

• Experience with Databricks and/or Genie

• Background in media, advertising, or marketing technology

• Experience building conversational interfaces or chatbot systems

• Familiarity with MLOps practices and model serving infrastructure

• Open-source contributions or personal projects demonstrating agentic AI work

Location:

DGS India - Bengaluru - Manyata N1 Block

Brand:

Dentsu

Time Type:

Full time

Contract Type:

Permanent

Requirements: