Leads the team this role joins
In-house, owns the opening
Put the opening up
From your own Gmail and LinkedIn.
About the role
This is a full-time AI engineering role in Hyderabad, India, focused on agentic AI enablement, for engineers with hands-on LLM application and production Python experience. The role builds reusable agent capabilities that run on PepsiCo's enterprise AI backbone across cloud and edge environments. It works alongside the AI Solutions Lead, turning domain workflows into reliable agent components.
What you’ll do
- Prototype, iterate and productionize agent modules for planning, tool use and task execution.
- Build and version agent assets such as prompts, policies, tool schemas and configs, with change logs.
- Tune agent latency and token use within set limits, including edge scenarios.
- Build an evaluation harness with regression suites, golden test sets and prompt or model comparisons.
- Track evaluation metadata and safety checks for hallucination, toxicity and policy compliance.
- Build or extend MCP-style connectors to internal data products and approved enterprise apps.
- Test integrations in sandbox environments with stakeholders.
- Add logging, telemetry, runbook notes and SLO alignment for owned components.
- Suggest small process improvements such as templates and prompt versioning conventions.
What they’re looking for
- A bachelor's or master's degree in computer science, AI, ML, data science, or equivalent experience.
- Hands-on experience building LLM applications with agents, tool use or prompting, and shipping Python to production.
- Production Python practice: testing, packaging and structured logging.
- Agent frameworks such as LangGraph, LangChain or CrewAI-style orchestration, plus tool calling and prompt versioning.
- Evaluation discipline: test sets, regression testing, offline metrics, A/B comparisons and failure taxonomies.
- Integration skills with APIs, auth and schema-based tools, with MCP-style interfaces preferred.
- Observability basics: correlation IDs, error analysis and latency instrumentation.
- Enough cloud knowledge to deploy and validate agents through platform pipelines.
Nice to have
- Ownership of components from prototype through to production-ready artifacts.
- A process improvement mindset that reduces rework.
- Clear communication of agent behavior and evaluation results to non-experts.
- Adaptability to changing model and tool constraints.
Skills
Questions about this role
Where is the job based?
Hyderabad, Telangana, India.
What kind of work is this?
Building agent components, evaluation test harnesses and MCP-style connectors to enterprise systems.
Is it full-time?
Yes.
Which skills are most important?
Production Python, agent frameworks such as LangGraph or LangChain, evaluation discipline and integration engineering.