Job Description

Generative AI is moving from assistants that answer questions to agents that do work. The Agentic AI Builder designs, builds and industrialises AI agents and multi-agent workflows for Sia's enterprise clients, and helps users adopt them. You will turn real business processes into agents that perform reliably in production.
This is agent engineering, not prompt writing or chatbot configuration. You will work with Forward Deployed Engineers, architects, consultants, and client teams, and you are expected to reason about orchestration, state, tools, evaluation, security, and cost, and to show with tests and measurements that an agent works.

Key Responsibilities:

- Work with consultants and business teams to identify agent use cases and translate business processes into transparent agent requirements and acceptance criteria.
- Design agent architectures for business workflows, deciding when a single agent, a multi-agent pattern or a deterministic workflow with LLM steps is the right choice.
- Implement orchestration, tool calling and control flow in Python using frameworks such as LangGraph, LangChain, the Claude Agent SDK or OpenAI SDK, or custom code where that is the better choice.
- Build tools and connectors to enterprise APIs and data sources, including MCP servers, with clear schemas, error handling and permissioning.




- Engineer context deliberately: retrieval over enterprise data, prompt and context design, and state and memory management across steps and sessions.
- Design human-in-the-loop checkpoints for approvals, escalation and exception handling.
- Create evaluation suites covering task success, tool-use correctness, regressions and adversarial cases, and use them to guide each release.
- Add observability, including traces, token and cost tracking and failure analytics, and optimise for reliability, latency and cost.
- Implement guardrails such as input and output validation, scoped credentials, prompt-injection defenses, and data-access controls.
- Work in agile, iterative cycles: estimate effort, flag risks and dependencies early, run solution demonstrations, and document agent behaviour and limits.
- Prepare agents for production with the FDE and architecture team, including deployment, monitoring, and operational handover.

Internal Contributions
As part of the AI Factory, you will also:

- Package reusable tools, connectors, and agent patterns that speed up later projects.
- Contribute to internal accelerators, demonstrations, and proposals.
- Share knowledge on agent design, evaluation, and guardrails with the wider engineering team.

📌 Agentic AI Builder (Mumbai)
🏢 Sia
📍 Mumbai