Senior AI Engineer

CittaAI; Hyderabad; On-site; INR 4,000,000 - 6,000,000; Full time

Job summary

CittaAI is seeking a Senior AI Engineer to take LLM-backed features from architecture to production, owning design choices around quality, latency, scale, and cost. You will lead end-to-end RAG pipelines, agent systems and the Python services that power them, with observability and evaluation driving ongoing health.

You will design backend services in Python with FastAPI and deliver interfaces in React/Next.js, while guiding team patterns and mentoring engineers to raise the overall quality and

Qualifications

• 6+ years in software or AI engineering with production Generative AI applications.
• Proficient in Python, REST APIs, FastAPI and distributed architectures.
• Hands-on experience with LLMs, RAG, embeddings, agents and tool calling.
• Familiarity with LangChain or equivalent orchestration framework.
• Experience with vector databases and semantic or hybrid search.
• Cloud-native development including CI/CD, containers, testing and debugging live systems.

Responsibilities

• Design and ship enterprise AI applications including copilots and knowledge assistants.
• Build RAG pipelines end to end: ingestion, chunking, embeddings, hybrid search, reranking and context construction.
• Develop agent systems with tool calling, state, memory, human-in-the-loop controls and failure recovery.
• Build Python backends with FastAPI and interfaces in React or Next.js.
• Define evaluation for groundedness, relevance, latency, reliability and cost.

Skills

Python

REST APIs

FastAPI

LLMs

LangChain

Vector DBs

CI/CD

Tools

Kubernetes

Docker

Helm

Terraform

GitHub Actions

Azure DevOps

pgvector

Pinecone

Weaviate

Job description

Senior AI Engineer who takes LLM-backed features from architecture to production and keeps them healthy — RAG pipelines, agent systems, the Python services behind them, and the evaluation and observability that prove they work. Senior IC scope: you own the design, make the trade-offs between quality, latency, scale, and cost, and set patterns other engineers build on.

Responsibilities

• Design and ship enterprise AI applications: copilots, knowledge assistants, and agentic workflows.
• Build RAG pipelines end to end — ingestion, chunking, embeddings, hybrid search, reranking, and context construction.
• Develop agent systems with tool calling, state, memory, human-in-the-loop controls, and failure recovery.
• Build the backend services and APIs behind them in Python and FastAPI, and the interfaces in React or Next.js.
• Define evaluation for groundedness, relevance, latency, reliability, and cost — not manual spot checks.
• Own deployment and operations: CI/CD, containers, cloud, observability, and incident response.
• Apply AI security and governance — auth/authz, prompt injection defense, sensitive-data handling, audibility.
• Lead design and code reviews, mentor engineers, and build reusable patterns.

Required Skills

• 6+ years in software or AI engineering, with production Generative AI applications you can walk us through.
• Strong Python, plus REST APIs, FastAPI, and distributed application architecture.
• Hands-on depth in LLMs, RAG, embeddings, agents, and tool calling.
• LangChain, LangGraph, or an equivalent orchestration framework.
• Vector databases and semantic or hybrid search.
• Cloud-native: Git, CI/CD, containers, automated testing, and debugging live systems.

Nice to have

• Production frontend with React and/or Next.js.
• Azure AI Search, Pinecone, Weaviate, or pgvector.
• LLMOps and evaluation tooling · observability platforms.
• Kubernetes, Docker, Helm, Terraform · Azure DevOps or GitHub Actions.

We are not looking for someone who is

• Stopping at the demo — prototypes that impress in a meeting but were never load-tested, evaluated, or handed to real users.
• Treating prompt tuning as the whole engineering job.
• Waiting for fully specified requirements; turning ambiguity into a shippable design is the work.
• Shipping without measurement — we expect evaluation harnesses and telemetry, not vibes.
• Uninterested in what happens after deploy: incidents, cost, latency, and drift are part of the role.
• Building alone — design reviews, code reviews, and mentoring are how the work compounds here.

Skills: