AI Native Prompt Engineer

EXPERIENCE LOCATION ROLE TYPE

GTE / Associate 6 12 months

(incl. Internships & Freelance) Gurugram Technical

About the Role

Prompt Engineers at Indus Towers are high-velocity builders who bridge frontier Large Language Models (LLMs), agentic orchestration frameworks, and real-world operational workflows. This is an early-career technical role for curious, fast-learning problem solvers who have spent their final semesters, internships, or freelance projects obsessively building with generative AI, agentic tools, and modern software runtimes.

You will not be writing theoretical papers or fine-tuning models from scratch. Instead, you will work on the frontlines alongside our Forward Deployed Engineers (FDEs), AI Catalysts, and the Head of Data Science & AI. Your mission is to rapidly test, iterate, and wire up structured prompts, multi-agent frameworks, tool-calling pipelines, and automated workflows that eliminate manual friction across thousands of telecom tower operations nationwide.

What You Will Do

Move from Idea to Working Code at High Velocity

Turn ambiguous process challenges, user stories, and Catalyst blueprints into functioning prompt prototypes and agentic micro-tools within hours or days.

Rapidly iterate on model prompting techniques-chain-of-thought, few-shot prompting, structured JSON outputs, and guardrails-to achieve high reliability on real-world edge cases.

Ship working MVPs with extreme speed, gathering early feedback from FDEs and domain leaders to continuously refine tool performance.

Maintain a relentless experimentation loop: test new open-source models, cutting-edge prompting patterns, and runtime libraries as soon as they drop.

Engineer Resilient Workflow Automations

Architect autonomous and semi-autonomous multi-agent workflows using modern frameworks (such as Mastra, Lang Chain, Llama Index, or CrewAI).

Design robust fallback logic, deterministic output validation (e.g., Pydantic / Zod schemas), and automated retries to prevent hallucination in mission-critical operations.

Automate end-to-end repetitive processes using a blend of Type Script/Python automation scripts, low-code orchestrators (n8n, Make), and programmatic webhooks.

Implement human-in-the-loop (HITL)



review gates so automated agents flag low-confidence decisions for manual review.

Integrate Enterprise Systems and Frontline Data

Build lightweight, robust API connectors and webhooks in Type Script and Python to link AI agents with enterprise databases, ERP/CRM records, and communication channels (Teams/Slack).

Implement Retrieval-Augmented Generation (RAG) pipelines over complex enterprise unstructured data, including SOPs, manuals, field tickets, and site inspection logs.

Query relational and non-relational databases (SQL, Vector databases) to feed clean, contextualized telemetry and site metadata directly into agent system prompts.

Ensure all agent interactions, API tokens, and data payloads strictly adhere to Indus Towers' cybersecurity, access control, and data privacy policies.

Build the Company Brain

Systematically document, version-control, and catalog system prompts, evaluation benchmarks, and agent tool definitions into a shared enterprise prompt repository.

Package proven agent skills, reusable Type Script/Mastra agent wrappers, and automated workflow recipes so other teams across Indus can deploy them instantly.

What We Are Looking For

Core Requirements

Relentless Curiosity & Prompt Craftsmanship: Intuitively understands how frontier models (OpenAI, Anthropic, Gemini, open-source LLMs) "think." Knows how to engineer bulletproof system prompts, handle token limits, enforce structured JSON schemas, and eliminate hallucinations.

High Velocity with Bias for Action: Prefers building a working prototype over endless debate. Quick to scaffold code, spin up an agent runtime, run evals, and show working software.

Hands-on Type Script & Modern AI Runtimes: Comfortable writing clean, modular code in Type Script and Node.js/Bun environments, leveraging agentic frameworks like Mastra to build deterministic workflows and tool-calling agents.

Data & Integration Fluency:



Comfortable querying databases via SQL, parsing messy JSON/XML payloads, reading REST API documentation, and connecting webhooks.

First-Principles Problem Solving: Breaks down complex, unstructured operational problems into clean logical steps that an AI agent or automated workflow can execute reliably.

Learning Agility & Adaptability: Unfazed by rapid shifts in the AI tooling landscape. Eager to master new libraries, frameworks, and agent paradigms overnight.

Experience and Education

Education: Bachelor's degree in Computer Science, Information Technology, Engineering, Data Science, or a related quantitative technical field (2024 / 2025 graduates welcome).

Professional Experience: 6 to 12 months of total hands-on experience building with up-to-date AI tools, agentic automation, and web APIs.

Note: High-impact college internships, freelance engagements, open-source contributions, and verified personal AI project portfolios count fully towards this experience.

Core Technical Stack:

Languages: Solid proficiency in Type Script / Java Script and Python.

Agentic Frameworks: Hands-on experience with modern agent frameworks such as Mastra, Lang Chain, Llama Index, or Auto Gen.

Prompting & Evals: Proven expertise in system prompt engineering, function/tool calling, structured output generation (Zod / Pydantic), and prompt evaluation techniques.

Databases & Vector Search: Working familiarity with SQL and vector databases / embeddings (e.g., Chroma, PGVector, Pinecone, Qdrant).

Workflows & Tools: Exposure to workflow automation platforms (n8n, Make), REST APIs, Git/Git Hub, and Docker fundamentals.

Portfolio Requirement: Active Git Hub repository, deployed demo, or live agent project demonstrating real-world LLM orchestration, structured tool use, or workflow automation.

How the Role Works

Dimension Details

Embedded with Forward Deployed Engineers (FDEs) and AI Catalysts

Delivers with Head of Data Science & AI, AI Platform Engineers, and Data Engineering teams

Owns System prompt engineering, agent tool creation, eval benchmarks, Type Script/Mastra workflow glue, and rapid MVP delivery

Operating principle Fast experimentation, strict output determinism; ship lightweight prototype

📌 Ai Prompt engineer (India)
🏢 Objectwin Technology India
📍 India