AI Harness Engineer

AI Harness Engineer ( Full Time )

Role Snapshot

Were hiring an experienced AI-native engineer to build the core systems behind Diligents AI agents. This is not a prompt-writing role. This is a software engineering role for someone who can design reliable AI systems: prompts context tools evals orchestration memory observability and production workflows.

Title: AI Harness Engineer

Experience: 5 years

Location: Jaipur

Role type: Full-time

Works with: Founders product growth and engineering teams

The Role

Youll work on the core agent architecture behind job will be to build the harness around frontier models: prompt engineering context engineering tool design agent orchestration evals memory model routing feedback loops and human-in-the-loop systems.

What Youll Do

Design and build AI agent workflows for marketing growth reporting creative analysis SEO/AEO lead intelligence and reputation monitoring.

Build production-grade harnesses around LLMs using prompts context tools memory structured outputs evals guardrails and human review loops.

Work on prompt engineering context engineering tool calling workflow decomposition and agent orchestration.

Build backend systems that connect with databases APIs CRMs ad platforms Slack WhatsApp dashboards and internal tools.

Design systems for ingesting cleaning storing and retrieving business context for AI agents.

Build evaluation pipelines to test prompt changes model changes agent quality regressions and edge cases.

Implement model routing strategies to balance quality latency and cost. Improve observability: logs traces prompt versions model outputs failure analysis and feedback loops.

Work closely with product and growth teams to convert messy business workflows into reliable AI systems.

Own engineering quality across architecture code testing debugging deployment and maintenance.

Who You Are

Must-haves

5 years of software engineering experience. Strong backend engineering and system design fundamentals.

Hands-on experience building with LLMs AI agents prompt engineering context engineering RAG tool calling or agent frameworks.

Strong software development fundamentals: clean code APIs databases queues testing debugging version control and deployment.

Experience with Python TypeScript or similar backend stacks.

Comfort working with Postgres Redis vector databases queues schedulersand production services.

Understanding of structured outputs JSON schemas retries validation error handling and deterministic workflows.

Ability to break ambiguous business workflows into clear technical systems.

High ownership and comfort working in an early-stage startup environment.

Nice-to-haves

Experience with OpenAI Anthropic LangGraph CrewAI LlamaIndex LangChain MCP or similar tools.

Experience building evals for AI systems.

Experience with tool servers workflow engines agent infrastructure or automation platforms.

Experience with marketing/adtech/growth tools such as Meta Ads Google

Ads GA4 HubSpot Zoho Shopify Slack WhatsApp or CRMs.

Experience with AWS GCP Azure Vercel Render Railway or similar cloud platforms.

Experience with observability tracing logging and production monitoring.

This Role Is Not For You If

You think AI engineering means only writing prompts.

You prefer demos over production systems.

You dont enjoy debugging messy edge cases.

You need perfectly defined requirements before starting.

You are uncomfortable owning systems end to end.

Interview Focus

Backend engineering and system design.

LLM/agent workflow experience beyond simple prompts.

Prompt engineering context engineering and harness engineering depth.

Ability to design evals feedback loops and observability for AI systems.

Ability to decompose complex workflows into reliable technical components.

Product thinking ownership and ability to work in ambiguity.

Required Skills:

BackendengineeringSystemdesignLLMsAIagentsPromptengineeringContextengineeringRetrievalaugmentedgenerationRAGToolcallingAgentorchestrationCleancodepracticesAPIdesignDatabaseintegrationQueuesystemsTestinganddebuggingVersioncontrolGitDeploymentpipelinesPythonNodejsTypeScript