Role Overview

You will own the complete Al backend end-to-end - from agent architecture and model integration to deployment, monitoring, and continuous improvement. You are the single owner of the Al stack: designing agents, building retrieval pipelines, optimizing costs,

deploying to production, and iterating based on real-world usage.

This role sits at the intersection of Al engineering, backend engineering, and DevOps. You won't just write prompts — you will architect systems, ship production code, debug production failures, and make cost-vs-quality tradeoffs that directly impact the business.

Design and scale a multi-agent orchestration system with 10+ specialized agents

Implement parallel execution (fan-out/fan-in), conditional routing, and shared state handling

Build current domain-specific agents with tailored retrieval, prompts, tools, and fallback strategies

Develop graph-based routing logic for dynamic task handling

1. LLM Integration & Optimization

Manage multi-model, multi-provider architecture Select models based on cost, latency, and accuracy tradeoffs

Build and refine prompt systems (few-shot, structured reasoning, domain tuning)

Optimize token usage, latency, and response quality

1. Retrieval & Knowledge Systems

Build hybrid search pipelines (vector + keyword + metadata filtering) Optimize chunking, indexing, embeddings, and re-ranking strategies

Ensure high-quality context retrieval for LLM pipelines

1.



Backend & Infrastructure

Develop scalable backend services for AI pipelines Implement streaming responses and async processing

Deploy and manage services using cloud infrastructure

Maintain CI/CD pipelines and production environments

1. Observability & Reliability

Build monitoring systems for latency, errors, and model performance Track hallucinations, failure cases, and edge scenarios

Implement fallback systems and reliability layers

1. Cost & Performance Optimization

Optimize inference costs across models and workflows Balance cost vs quality vs latency in production systems

Continuously improve system efficiency.

Strong experience with backend development (Python preferred)

Hands-on experience with LLMs and AI systems in production

Experience with vector databases and retrieval systems

Understanding of distributed systems and async processing

Familiarity with cloud platforms (AWS/GCP/Azure)

Nice to Have

Experience with multi-agent frameworks

Knowledge of prompt engineering at scale

Experience with real-time AI applications

Exposure to monitoring and observability tools

Strong ownership mindset

Ability to work across AI, backend, and infrastructure

Focus on shipping production-ready systems

Practical problem-solving over theoretical approaches.

Pay: ₹500,000.00 - ₹600,000.00 per year

Work Location: In person

📌 Al Engineer - Full Stack Al Backend & Infrastructure (Pune)
🏢 Cybic Technologies
📍 Pune