Generative AI & Machine Learning | Google Cloud Platform | Federal Program
Location: Remote (United States) — occasional travel may be required Employment Type: Full-Time (Contract) Rate: $100 – $150 per hour (commensurate with experience) Level: Mid-Senior / SME (3–6+ years) Citizenship: U.S. Citizenship required — no exceptions, no visa sponsorship, no C2C through third-party visa holders Clearance: Must be able to obtain and maintain Minimum Background Investigation (MBI)
About the Role
Lanthos Tech is seeking a GCP AI Engineer (SME) to design, build, and scale a production-grade AI research system on Google Cloud Platform.
This role sits at the intersection of applied AI, software engineering, and federal compliance. You will build Retrieval-Augmented Generation (RAG) pipelines, deploy models through Vertex AI, and integrate Gemini APIs — all within a FedRAMP-authorized, ATO-governed cloud environment. It suits an engineer who is equally comfortable writing production Python, architecting cloud infrastructure, and reasoning about model behavior, latency, cost, and auditability.
What You'll Do
- Design, build, and deploy machine learning and generative AI pipelines on Vertex AI , including model training, fine-tuning, evaluation, and serving.
- Build RAG architectures using vector search (e.g., Vertex AI Vector Search) to ground LLM outputs .
- Integrate Gemini APIs and other foundation models into mission applications and internal analyst tooling.
- Develop agentic workflows and multi-step orchestration using frameworks such as LangChain, LangGraph, or CrewAI .
- Build and deploy agents using Google's Agent Development Kit (ADK) and integrate them with Gemini Enterprise for enterprise-grade deployment, access control, and governance.
- Operate agents within managed runtime environments (e.g., Vertex AI Agent Engine ), handling session state, tool-calling, and scaling of long-running agent workloads.
- Architect and maintain scalable, secure cloud infrastructure across GCP services ( BigQuery, Cloud Run, Pub/Sub, Cloud Functions, GKE ), using FedRAMP-authorized services and government-approved regions .
- Implement MLOps practices: CI/CD for models, automated retraining, monitoring, drift detection, evaluation harnesses, and rollback strategies.
- Design systems that meet federal security and privacy obligations.
- Support the Authority to Operate (ATO) process by producing architecture artifacts, data-flow diagrams, and control evidence in partnership with the security and compliance lead.
- Write clean, well-tested, production-quality Python .
- Collaborate with data scientists, product managers, platform engineers, and government stakeholders to translate mission requirements into technical solutions.
- Monitor and optimize model performance, latency, and inference cost in production.