AI Agent Engineer

Blue Machines AI; Mumbai; On-site; INR 350,000 - 650,000; Full time

Job summary

Blue Machines AI in Mumbai and Chennai is seeking an AI Agent Engineer with 5+ years of hands-on experience shipping an AI feature used by real users. You will design, build, integrate, and debug the agent end-to-end, including conversation prompts, tool-calling logic, and basic integrations.

You will read logs to diagnose issues, perform experiments to improve latency and cost, and communicate clearly with clients about progress and timelines.

Qualifications

• 5+ years of hands-on experience shipping an LLM/AI feature used by real users.
• Ability to read logs and reason whether an issue is prompt, platform, or data.
• Comfortable pulling data in SQL, Sheets, or Python.
• Fluent in English and Hindi; willing to travel for client workshops.
• Able to explain technical issues clearly to clients and set expectations.

Responsibilities

• Conversation design, prompts, guardrails, and tool-calling logic.
• Basic integrations (webhooks, CRM writes, callbacks) your agent depends on.
• Debugging to identify root cause (prompt, platform, or data) rather than guessing.
• Evaluate metrics: hallucination rate, completion rate, latency, and cost.
• Run experiments on agent funnels with defined success metrics.
• Conduct client-facing calls: explain issues and share timelines.

Skills

Fluent English

Fluent Hindi

SQL

Python

Log analysis

Client communication

Tools

SQL

Python

Sheets

Job description

AI Agent Engineer

Location: Mumbai (1 position) , Chennai (2 positions)

Experience: 5+ years

Blue Machines builds production Voice AI for Indias largest enterprises — millions of live conversations in Hindi, Hinglish, and regional languages. You design, build, integrate, and debug the AI agent end-to-end. You are delivery for your agent, not a partner handing off to it.

You’ll Own

• Conversation design, prompts, guardrails, and tool-calling logic


• Basic integrations (webhooks, CRM writes, callbacks) your agent depends on


• Debugging: given a bad call, prove whether its prompt, platform, or data — not guess


• Evaluation and iteration: hallucination rate, completion rate, latency, cost


• Experiments on your own agents funnel (opening line, retry cadence, sequencing) with a defined success metric


• Client-facing calls on your agent: explain what broke, what youre doing about it, and set realistic timelines

Must Have

• 5+ years of experience, with hands‑on work shipping an LLM/AI feature real users touched — chatbot, RAG, or agent (voice a plus, not required)


• Can read a log or trace and reason whether an issue is prompt, platform, or data


• Comfortable pulling your own data in SQL, Sheets, or Python


• Fluent English and Hindi


• Willing to travel for client workshops and live-call reviews


• Comfortable in front of clients — can explain a technical issue in plain terms and manage expectations without becoming defensive

Good to Have

• Tamil, Telugu, Kannada, Malayalam, Marathi, or Gujarati


• BFSI/lending exposure

Non-Ideal Fit

• Ship-and-forget builders uninterested in monitoring or RCA


• Spec-writers who dont build

Process

Screen → live take-home (build + deploy a working voice agent) → deep-dive on what you built → final conversation. The take-home is the real gate on voice-specific skill — not the resume.

Requirements: