Job Summary

We are seeking an Agentic Software Developer at MTS+ or equivalent scope to build intelligent automation for IOQ, AMD Fleet Services. This role will design and engineer agent-assisted workflows that improve alert intake, ticket triage, evidence collection, knowledge retrieval, procedure execution, operational reporting, and safe remediation across compute, storage, networking, identity, and cloud services.

The developer will convert operational intent into reliable software. The focus is not on demonstrations alone, but on governed, observable, testable, and maintainable capabilities that operate within approved access, change, data-handling, and human-approval boundaries.

Key Responsibilities

Agent Development 

• Architect and develop agentic workflows in Python using Google ADK — agents and sub-agents, tool definitions, session and state management, callbacks, multi-agent delegation, and artifact handling — packaged for repeatable, version-controlled deployment. 
• Build workflows for alert enrichment, Jira intake, classification, evidence collection, knowledge retrieval, recommended actions, handoff preparation, and operational reporting. 
• Integrate approved AI models and tools with monitoring, ticketing, documentation, inventory, and automation platforms through secure APIs and controlled service identities. 
• Engineer deterministic orchestration, validation rules, confidence handling, human approval gates, stop conditions, rollback, and audit trails for operational use cases. 

 

 

 

Fullstack Delivery 

• Design and build the fullstack surface for agentic services: backend APIs and service layer (FastAPI or equivalent), plus the operator-facing web application where approval gates, evidence review, override, and audit history are exposed. 
• Own the end-to-end path from data model and API contract through authentication, state management, front-end implementation, and deployment — including responsiveness, accessibility, and behavior under partial failure or streaming/long-running agent responses. 

 

Evaluation, Telemetry & Reliability 

• Develop evaluation suites using representative and historical incidents to measure accuracy, completeness, repeatability, safety, and escalation quality. 
• Implement telemetry for prompts, tool calls, latency, failures, costs, outcomes, user overrides, and operational impact while protecting confidential data. 
• Support production incidents involving agentic services and drive root-cause correction, reliability improvement, and reduction of false or inconsistent outcomes. 

 

Standards, Security & Enablement 

• Partner with L1/L2/L3 domain owners to codify SOPs and runbooks into reusable automation components without obscuring ownership or change-control requirements. 
• Create version-controlled software, prompts, policies, schemas, tests, deployment pipelines, documentation, and operational support procedures. 
• Apply secure software development practices, least privilege, secrets management, input/output validation, dependency controls, and appropriate data retention. 
• Establish reusable patterns for agent design, tool integration, evaluation, front-end operator experience, and production readiness across Fleet Services. 

Skill Requirements

Candidates must demonstrate hands-on, production-grade experience in each area below. 

• Python (expert level). Strong software engineering experience in Python, including API development, type hints, asynchronous and concurrent systems, packaging, automated testing, and production support of services you wrote. 
• Google Agent Development Kit (ADK). Demonstrated experience building and shipping agents with ADK — agent and tool definition, session/state handling, callbacks, multi-agent orchestration, artifact management, and ADK's evaluation tooling. Be prepared to discuss a specific agent you built, what broke in production, and how you fixed it. 
• AI fundamentals. Working command of the mechanics beneath the frameworks: prompt and context construction, context-window and token budget management, embeddings and retrieval-augmented generation (chunking, indexing, retrieval quality), tool/function calling, structured output and schema validation, temperature and sampling behavior, hallucination and model failure modes, and how to design an evaluation that actually detects regression. 
• Fullstack engineering. End-to-end delivery of web applications: TypeScript/JavaScript with a modern front-end framework (React or equivalent), REST or GraphQL API design, authentication and authorization, relational and/or vector data stores, and deployment through CI/CD. You should be able to build the operator UI for your own agent without waiting on another team. 
• Enterprise integration. Experience integrating with platforms such as Jira, Confluence, Git, monitoring systems, automation controllers, or cloud services through authenticated APIs. 
• Secure AI application design. Strong understanding of prompt injection, privilege boundaries, secrets, data classification, auditability, model failure modes, and safe tool-execution patterns. 

Other Requirements

• A second production language among Go, Java, C++, or Rust. 
• Broader agent-framework exposure (LangGraph, Semantic Kernel, MCP-based tool servers) and the judgment to explain where each fits relative to ADK. 
• Google Cloud AI stack beyond ADK — Vertex AI, Gemini APIs, Agent Engine, or comparable managed agent runtimes. 
• Infrastructure automation and configuration tooling such as Ansible, Terraform, shell, or CI/CD systems. 
• SRE practices: incident management, observability, runbooks, SLIs/SLOs, and production change control. 
• Ability to influence architecture across teams and communicate tradeoffs to both technical and operational stakeholders. 
• GPU/HPC infrastructure, Azure AI services, or on-premises model deployment. 

ACADEMIC CREDENTIALS

Bachelor’s or Master’s degree in Computer Science, Computer Engineering, Software Engineering, or a related technical field preferred; equivalent relevant experience considered.