AI Platform and Solutions Engineer

Invent the future with us.

Ampere is a semiconductor design company for a new era, leading the future of computing with an innovative approach to CPU design focused on high-performance, energy efficient AI compute.

As an AI Platform & Solutions Engineer, you will build and operate Ampere’s internal AI platform and the user-facing tools that sit on top of it. You’ll deliver secure AI applications (chat/agent/workflow UIs), develop integrations that connect LLMs to enterprise systems, and run core platform components (vector/RAG services, model access, deployments, monitoring). Your work will directly improve developer productivity, knowledge discovery, and operational automation across the company while protecting Ampere IP.

This is a builder/operator role: ship features, wire them into real systems, and keep them reliable.

AI Platform work at Ampere is focused on enabling every employee with practical, secure AI capabilities. The team builds foundational platform services and high-impact internal tools from the ground up, partnering with engineering, manufacturing, and operations to turn ambiguous workflows into deployed systems. You’ll operate close to production realities: identity, permissions, observability, cost controls, and uptime matter as much as model quality.

AI Infrastructure & Platform Operations

• Core stack:
  Python + FastAPI, Svelte/Svelte Kit + Type Script, Azure, Microsoft Entra.
• AI infrastructure management:
  Own and evolve core components like vector/RAG services, model access, model serving environments (cloud and on-prem where applicable), and underlying compute.
• Containerization & deployment:
  Package and deploy services with Docker, primarily to Azure VMs (and adjacent Azure services as needed).
• Observability & reliability:
  Implement logs/metrics/traces using Azure Monitor/Application Insights, build dashboards, and configure actionable alerts.
• Secrets, config, and access:
  Manage secrets via Key Vault, use managed identities, and enforce least-privilege access patterns.
• CI/CD:
  Build pipelines (Git Hub Actions and/or Azure Dev Ops) for build/test/deploy across environments.

• Backend development:
  Build and extend FastAPI services (endpoints, adapters, background jobs, structured payloads, pagination, and error handling).
• Model Context Protocol (MCP) / tool integrations:
  Design and maintain MCP-style tools and connectors that link models/agents to enterprise systems like Jira/Confluence, ticketing systems, knowledge bases, Net Suite/manufacturing tools, and internal services.
• RAG implementation:
  Build practical RAG foundations: chunking, embeddings, retrieval filters/metadata, batch upserts, and index maintenance across vector/search systems.
• LLM integration:
  Integrate with LLM APIs (Azure Foundry/Google Vertex AI) including streaming and tool/function calling.

• Internal AI web apps:
  Own user-facing web UIs (primarily Svelte/Svelte Kit + Type Script) with strong fundamentals around routing, SSR, state, performance, and accessibility.
• Real-time AI UX:
  Implement streaming responses via SSE/Web Sockets, resilient chat/task interfaces, incremental rendering, retries, and degraded-mode handling.
• Agents and workflow automation:
  Build multi-step agents and workflows with tool calls, approvals, progress events, audit trails, and safe failure modes.
• Solution delivery:
  Ship tools that improve day-to-day work, including internal chat/agent apps, workflow automation, and developer productivity integrations (e.g., VS Code, Open WebUI, external AI tools where appropriate).

• Authentication and authorization:
  Implement Microsoft Entra  for web apps/APIs using OIDC/OAuth2 (MSAL), validate tokens (JWT/JWKS), and enforce scopes/RBAC.
• Data protection and auditability:
  Ensure safe enterprise use of LLMs with clear permission boundaries, logging/audit trails, and secure handling of sensitive data.

• Partner with internal teams (engineering, manufacturing, ops) to gather requirements, translate them into working systems, document…