GCP Agentic AI & Integration Engineer

U.S. citizenship is required by the federal engagement supporting this role. Candidates must reside in and perform all work from within the United States. Work is remote, although occasional approved travel may be required. Visa sponsorship is not available.

Lanthos is hiring a GCP Agentic AI & Integration Engineer to implement secure agents powered by Gemini models through Vertex AI, along with their tools and enterprise integrations, in a regulated public-sector environment.

This is a junior-to-mid-level, implementation-focused role. You will build agent workflows and the APIs that connect them to approved enterprise systems within established architecture and security standards. You will collaborate with senior AI, platform, data and security engineers rather than independently setting the enterprise agent strategy.

What you’ll do

• Build Python services that call Gemini models through Vertex AI APIs.
• Implement agent workflows, tools, session state, routing and controlled multi-step execution using Google Agent Development Kit or another approved orchestration framework.
• Deploy or integrate agent services using Vertex AI Agent Engine or Gemini Enterprise where appropriate.
• Connect agents to approved enterprise content, search services and internal APIs using grounded, least-privilege patterns.
• Build and maintain REST APIs and tool interfaces with authentication, authorization, quotas, rate limits and resilient error handling.
• Deploy containerized agent and integration services to Cloud Run or another approved Google Cloud runtime.
• Add evaluations, guardrails, structured logging, tracing and operational alerts for tool use and agent failures.
• Create automated tests for agent behavior, API contracts, authentication and failure recovery.
• Maintain tool specifications, interface documentation, deployment notes and operational runbooks., * $70.00-$90.00 per hour, based on relevant experience and final assigned scope.
• Independent-contractor engagement, anticipated up to 40 hours per week based on project needs and approved work.
• Direct engagement with the individual or a candidate-owned corporation may be considered.
• This contract engagement does not include company-sponsored employee benefits. No bonus, commission or equity compensation is currently offered.
• Third-party staffing submissions are not accepted.
• Applications are expected to remain open through August 21, 2026. The deadline may be extended if the position remains open.

Interview process

• Initial conversation covering eligibility, availability and engagement expectations.
• Practical technical interview focused on agent design, Gemini models through Vertex AI APIs, authentication, Cloud Run and an AI integration the applicant personally delivered.
• Team conversation covering troubleshooting, collaboration and regulated-program fit.

Lanthos LLC is an equal opportunity employer. The citizenship and suitability requirements above are imposed by the federal engagement supporting this position.

• Two or more years of backend, AI-application or integration-engineering experience, including at least one year of hands-on Google Cloud work.
• Strong practical Python skills plus experience with REST APIs, JSON, authentication and service-to-service communication.
• Hands-on experience personally building an LLM agent, tool-calling workflow or AI integration using Gemini on Vertex AI.
• Experience deploying a containerized API, agent or integration service on Google Cloud.
• Familiarity with Git, automated testing and CI/CD workflows.
• Working knowledge of IAM, secrets management, logging and monitoring.
• A current Google Cloud Associate Cloud Engineer, Professional Cloud Developer or Professional Machine Learning Engineer certification.
• Ability to troubleshoot integrations and explain personal implementation work clearly.

Preferred qualifications

• Hands-on experience building agent workflows with Google Agent Development Kit or another production orchestration framework.
• Experience deploying agents to Vertex AI Agent Engine or integrating solutions with Gemini Enterprise.
• Experience with Vertex AI Search, Vector Search, Document AI or grounded enterprise retrieval.
• Google Cloud Professional Cloud Developer or Professional Machine Learning Engineer certification.
• Experience with Cloud Run, IAM, Secret Manager, API Gateway, Apigee, logging or monitoring.
• Experience with OAuth 2.0, OIDC, JWTs, mTLS or workload identity.
• Familiarity with enterprise content, document or data-storage systems.
• Experience in a federal, public-sector or regulated environment.

Federal program requirements

• Must be a U.S. citizen.
• Must reside in and perform all work from within the United States.
• Must be able to obtain and maintain the required MBI/Public Trust suitability determination.
• Must be willing to complete required fingerprinting, identity verification, security and privacy training and applicable compliance checks., * Are you currently located in the United States and able to perform all work from within the United States? (Yes/No)
• Are you willing and able to complete an MBI/Public Trust suitability process, including fingerprinting and required compliance checks? (Yes/No)

$70 - $90 an hour - Contract, * Are you available for a contract engagement anticipated up to 40 hours per week and comfortable proceeding within the posted $70-$90/hour range? (Yes/No)

• How many years of professional backend, AI-application or integration-engineering experience do you have? Enter a number.
• Do you currently hold an active Google Cloud Associate Cloud Engineer, Professional Cloud Developer or Professional Machine Learning Engineer certification? (Yes/No)
• Provide the certification name, expiration date and a public credential link or credential ID.
• Which of the following have you used hands-on? List all that apply: Gemini on Vertex AI APIs; Google Agent Development Kit; Vertex AI Agent Engine; Gemini Enterprise; tool/function calling; agent session state; Vertex AI Search/Vector Search; Cloud Run; OAuth/OIDC/JWT; or none.