GenAI & Agentic AI Engineer

Brandsmashers Tech; Bhopal; On-site; INR 900,000 - 1,800,000; Full time

Job summary

Brandsmashers Tech seeks an engineer to build LLM-powered products and multi-agent systems that take real actions safely. You will develop retrieval-augmented pipelines over client data, design auditable agents, and implement guardrails before shipping features.

The role emphasizes writing robust evaluation sets, integrating LLM APIs with business systems, and monitoring cost, latency, and quality in production. Strong Python or TypeScript skills are essential.

Qualifications

• Proficient in Python or TypeScript with production experience.
• Experience building with LLM APIs and embeddings.
• Familiarity with vector databases and retrieval quality.
• Understanding of prompt design, tool calling and structured outputs.
• Ability to measure quality and performance beyond eyeballing.

Responsibilities

• Build retrieval-augmented generation pipelines over client data.
• Design agents and tools with narrow, auditable permissions.
• Write evaluation sets and guardrails before features ship.
• Integrate LLM APIs with business systems and human handoffs.
• Track cost, latency and quality in production.

Skills

Python

TypeScript

LLM APIs

Embeddings

Vector databases

Tools

LangGraph

LlamaIndex

OpenAI API

Job description

Build LLM-powered products and multi-agent systems that take real actions safely: support agents, copilots, retrieval over company knowledge and orchestrated workflows.

What you'll do

• Build retrieval-augmented generation pipelines over client data
• Design agents and tools with narrow, auditable permissions
• Write evaluation sets and guardrails before features ship
• Integrate LLM APIs with business systems and human handoffs
• Track cost, latency and quality in production

What you'll bring

• Strong Python or TypeScript engineering skills
• Hands-on experience building with LLM APIs and embeddings
• Experience with vector databases and retrieval quality
• Understanding of prompt design, tool calling and structured outputs
• A habit of measuring quality rather than eyeballing it

Nice to have

• LangGraph, LlamaIndex or similar frameworks
• Fine-tuning or small open models
• Voice or multimodal interfaces