This role will be based in Sao Paulo or Brasilia.
Responsibilities
- Architect and manage Large Language Model (LLM) deployments across on-premises (NVIDIA/AMD) and cloud (cloud computing platform, Google Cloud platform (GCP) environments. Audit multi-agent orchestration, agent construction, and vector databases to map data flows and enforce privilege boundaries.
- Bridge local high-compute clusters and cloud AI services while maintaining a consistent security posture.
- Use Docker and Kubernetes to orchestrate scalable inference and training environments, optimizing Graphics Processing Unit (GPU) utilization and resource isolation.
- Protect model weights, secure data ingestion, and harden inference endpoints across the Machine Learning operations (MLOps) life-cycle.
- Investigate and mitigate AI-specific threats (e.g., prompt injection, jailbreaking, data poisoning). Map testing findings to MITRE ATLAS, OWASP for LLMs, and STRIDE models.
Minimum qualifications:
- Bachelor's degree or equivalent practical experience.
- 6 years of experience with cloud native architecture in a customer-facing or support role.
- Experience with cloud engineering, on-premise engineering, virtualization, or containerization platforms.
- Experience engaging with, or presenting to, technical stakeholders or executive leaders.
- Experience in cloud migration or transformation projects in the Public Sector.
- Experience in a technical pre-sales role (e.g., partner engineer, partner solution architect, solution architect, sales engineer, customer engineer, field application engineer).
Preferred qualifications:
- Experience deploying AI models on air-gapped or on-premises high-performance computing (HPC) systems.
- Experience with Large Language Model (LLM) deployment frameworks such as vLLM, NVIDIA Triton, or Ollama and agent development.
- Experience building long-term technical strategies to uncover new workloads with customers.
- Familiarity with cloud-native AI services (e.g., cloud computing platform, Google Vertex AI).
- Knowledge of Open Worldwide Application Security Project (OWASP) for LLMs or other security frameworks.