Senior AI Engineer

This is the senior build role for Nexperias agent engineering practice in Bangalore. You own the architecture of the agents themselves reasoning planning tool-calling and memory and you own the surfaces people actually use them through: chat UI IDE extensions CLI tools and the APIs behind them.

You work closely with the Integration Architect on how agents connect into IDEs and internal systems and with the MLOps/eval function on the quality bar an agent has to clear before it ships.

This is a hands-on role. You are still writing and reviewing code still debugging orchestration failures when something breaks in production and still accountable for what ships not a role that manages from a distance.

Architect and orchestrate agents

• Design agent architecture: reasoning loops planning tool-calling and memory/context management.

• Build multi-step orchestration: task decomposition tool selection and error recovery.

• Partner with the Integration Architect on connecting agents to IDEs and internal systems.

Ship end-to-end agent products

• Own the build of user-facing surfaces for Copilot-style products: chat UI IDE extensions and CLI tools.

• Build the backend APIs that connect those surfaces to agent and LLM services including streaming responses session state and context handling.

• Ship integrations into developer tools (VS Code extension API LSPs browser extensions) and own deployment of both the frontend and backend components you build.

Set the bar and take releases live

• Review code and agent designs from the two Junior AI Engineers and mentor them toward independent ownership.

• Define the agent quality bar with the MLOps/eval function before anything ships and work with that function on the CI/CD monitoring versioning/rollback and observability an agent needs once its live.

Enable the Spokes

• Own the governance bar for citizen-built agents what safe to ship means for security data access cost and quality in partnership with security/compliance and the Integration Architect.

• Publish and maintain the standards prompting evals architecture patterns that spoke teams build against.

• Own the design of the self-service component catalog spokes build on top of and the support model office hours an embedded rotation or a ticket queue for citizen developers using it.

We have kept this list to what the role truly requires. If this role excites you and you meet most but not all of these points we encourage you to apply.

Core requirements

• 1-6 years hands-on building LLM or agent systems.

• Strong Python and/or TypeScript; comfortable owning a service end to end.

• Hands-on with agent frameworks or SDKs (Claude Agent SDK/MCP LangGraph AutoGen) or custom orchestration.

• Has shipped an agentic system to production not just a prototype.

• Solid grasp of traditional machine learning/AI (e.g. classification regression clustering feature engineering) in addition to LLM/agent-based approaches..

• Comfortable designing REST GraphQL and streaming/WebSocket APIs and familiar with LLM streaming APIs (SSE/WebSockets) and token-by-token rendering.

• Familiarity with enterprise AI/copilot platforms and cloud-native model ecosystems Microsoft Copilot and Copilot Studio Azure AI Foundry (model catalog deployments and orchestration) and AWS model configuration (Bedrock model access inference profiles/settings) is a plus.

Valuable extras

• Strong in React/TypeScript for frontend work and or Python for backend work.

• Experience with CI/CD tooling (GitHub Actions Jenkins or similar) cloud infrastructure (AWS Azure or GCP) and containers (Docker Kubernetes).

• Observability tooling Datadog Prometheus/Grafana or LLM-specific tools like LangSmith or Arize and familiarity with LLMOps practices: prompt versioning eval pipelines cost tracking.

• Working knowledge of auth and session management and awareness of security and compliance considerations for AI systems.

• Experience in the manufacturing industry preferably semiconductor is a plus.

• Experience or interest in developer enablement or platform engineering documentation SDKs or internal tooling built for non-specialist users not just for your own team.

• Real ownership: you set the architecture and the quality bar for agents that ship to real users not just a reference design others build from.

• A mentoring role with teeth: two Junior AI Engineers are counting on your reviews and Nexperia measures you on their growth as much as on your own output.

• Youre measured not only on what the Hub ships but on how many spoke teams successfully ship agents using what youve published.

• Global exposure from Bangalore: you work with the AI and Data Centers of Enablement the Integration Architect platform engineering security and compliance with visibility up to the CIO and IT Management Team.

• A team that values structured thinking intellectual honesty about trade-offs and good documentation over heroics.

Talent acquisition based on Nexperia vacancies is not appreciated. Nexperia job adverts are Nexperia copyright material and the word Nexperia is a registered trademark.

D&I Statement

As an equal-opportunity employer Nexperia values diversity not just because it is the right thing to do but because diverse teams perform better. We are dedicated to being inclusive and a proof point of this dedication is that we were the main partner of the very first Dutch Paralympic Team NL House during the Paris 2024 Paralympic Games. Our recruitment process is inclusive and accessible to all and we consider all applicants fairly as well as providing a safe work environment and reasonable adjustments where requested.

In addition we offer our colleagues the possibility to join employee resource groups such as the Pride Network Group or global and local Womens groups. Nexperia is committed to increasing women in management positions to 30% by 2030.