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 clear 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.

Education and experience

• Bachelor’s or Master’s degree in Computer Science, Engineering or a related field, or equivalent practical experience.
• 2 to 6 years of software or AI engineering experience.
• At least one agent or LLM workflow delivered beyond proof of concept, with evidence of how it was tested and measured.
• Professional English, written and spoken.
  Technical skills (required)
• Strong Python and software engineering fundamentals: testing, packaging, typing and asynchronous programming.
• Hands-on experience building agents with tool calling and orchestration using at least one agent framework or SDK (LangGraph, LangChain, Claude Agent SDK, OpenAI SDK or equivalent). Experience with every named framework is not expected.
• Understanding of agent state, memory and control flow, including retries, timeouts and failure recovery.
• Practical experience with RAG and vector search, and with context and prompt engineering as part of a larger system.
• Experience evaluating LLM and agent behaviour with test sets, automated checks or LLM-as-judge methods.
• Familiarity with LLM observability and tracing, and with tracking cost and latency.
• Experience integrating agents with enterprise APIs, databases or SaaS applications.
• Awareness of agent-specific risks, including prompt injection, excessive tool permissions and data leakage, and how guardrails address them.
• Cloud deployment fundamentals: containers, environment configuration and CI/CD.
  Consulting and collaboration
• Ability to explain technical decisions and limits clearly to engineers, architects, and non-technical colleagues. Comfort running solution demonstrations and gathering feedback from business users.
• Ownership of the components you build and the ability to work with geographically distributed teams.
  Good to have
• Working knowledge of the Model Context Protocol (MCP), including building or hosting an MCP server.
• Experience with multi-agent patterns such as planner and executor, supervisor, or handoff designs.
• Experience with automation platforms such as n8n or Microsoft Copilot Studio alongside code-first approaches.
• Exposure to model routing, fine-tuning, or model selection trade-offs.
• FastAPI or other backend service experience; Kubernetes and infrastructure as code basics.Experience in a regulated sector or collaboration with international teams and clients.

What We Offer 

• The opportunity to work on production AI for global enterprise clients, as part of a growing AI engineering capability.

• The chance to learn through real client implementations and to build reusable agent assets for the AI Factory.

• Access to Sia’s leadership and training programmes as you grow.

• A collaborative, multidisciplinary team working across varied projects and industries.

Position based in Mumbai (onsite)

Equal Opportunity

Sia is an equal opportunity employer. Decisions on hiring, promotion, remuneration and discipline are made solely on performance, competence, conduct and business needs.

Sia is an equal opportunity employer. All aspects of employment, including hiring, promotion, remuneration, or discipline, are based solely on performance, competence, conduct, or business needs. 

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 clear 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.

Education and experience

• Bachelor’s or Master’s degree in Computer Science, Engineering or a related field, or equivalent practical experience.
• 2 to 6 years of software or AI engineering experience.
• At least one agent or LLM workflow delivered beyond proof of concept, with evidence of how it was tested and measured.
• Professional English, written and spoken.
  Technical skills (required)
• Strong Python and software engineering fundamentals: testing, packaging, typing and asynchronous programming.
• Hands-on experience building agents with tool calling and orchestration using at least one agent framework or SDK (LangGraph, LangChain, Claude Agent SDK, OpenAI SDK or equivalent). Experience with every named framework is not expected.
• Understanding of agent state, memory and control flow, including retries, timeouts and failure recovery.
• Practical experience with RAG and vector search, and with context and prompt engineering as part of a larger system.
• Experience evaluating LLM and agent behaviour with test sets, automated checks or LLM-as-judge methods.
• Familiarity with LLM observability and tracing, and with tracking cost and latency.
• Experience integrating agents with enterprise APIs, databases or SaaS applications.
• Awareness of agent-specific risks, including prompt injection, excessive tool permissions and data leakage, and how guardrails address them.
• Cloud deployment fundamentals: containers, environment configuration and CI/CD.
  Consulting and collaboration
• Ability to explain technical decisions and limits clearly to engineers, architects, and non-technical colleagues. Comfort running solution demonstrations and gathering feedback from business users.
• Ownership of the components you build and the ability to work with geographically distributed teams.
  Good to have
• Working knowledge of the Model Context Protocol (MCP), including building or hosting an MCP server.
• Experience with multi-agent patterns such as planner and executor, supervisor, or handoff designs.
• Experience with automation platforms such as n8n or Microsoft Copilot Studio alongside code-first approaches.
• Exposure to model routing, fine-tuning, or model selection trade-offs.
• FastAPI or other backend service experience; Kubernetes and infrastructure as code basics.Experience in a regulated sector or collaboration with international teams and clients.

What We Offer 

• The opportunity to work on production AI for global enterprise clients, as part of a growing AI engineering capability.

• The chance to learn through real client implementations and to build reusable agent assets for the AI Factory.

• Access to Sia’s leadership and training programmes as you grow.

• A collaborative, multidisciplinary team working across varied projects and industries.

Position based in Mumbai (onsite)

Equal Opportunity

Sia is an equal opportunity employer. Decisions on hiring, promotion, remuneration and discipline are made solely on performance, competence, conduct and business needs.

Sia is an equal opportunity employer. All aspects of employment, including hiring, promotion, remuneration, or discipline, are based solely on performance, competence, conduct, or business needs.