Key Facts
Location: Buenos Aires, Argentina
Engagement: Full-time
Team: ML Inference Platform
What You’ll Do
- Design and maintain the systems that bridge AI model development with production inference.
- Manage containerized workloads on Kubernetes and GCP, optimizing for NVIDIA GPU, memory, and network performance.
- Integrate and adapt model-serving frameworks such as vLLM, Triton, or TGI.
- Implement safety mechanisms for model releases, including shadow serving, canary rollouts, and rapid rollback procedures.
- Develop infrastructure for benchmarking latency, throughput, and cost under real-world traffic conditions.
- Standardize the lifecycle for model artifacts, including versioning, promotion, and deployment across environments.
- Enhance observability through telemetry, structured logging, and alerting to monitor production behavior.
- Improve compute efficiency and autoscaling strategies to optimize cost and performance.

Requirements
- 6+ years of professional software engineering experience.
- Proficiency in writing maintainable production code using Python,



Go, or similar backend languages.
- Proven background in operating high-throughput distributed systems or data/ML infrastructure.
- Practical knowledge of Linux, Kubernetes, containerization, and CI/CD operations.
- Ability to analyze performance bottlenecks across compute, memory, and concurrency.
- Strong operational mindset regarding system resilience, failure modes, and observability.
- Experience collaborating with cross-functional teams to transition AI capabilities into production services.

Skills & Tools
- Languages: Python, Go
- Infrastructure: Kubernetes, GCP, NVIDIA GPU
- Frameworks: vLLM, Triton, TGI
- Operations: CI/CD, Observability, Telemetry, Benchmarking, Distributed Systems

Practical Notes
- This is a new team focused on building core inference infrastructure rather than research or generic MLOps support.
- Dialpad offers competitive salary and comprehensive benefits.

📌 AI Systems Engineer (Argentina)
🏢 Dialpad
📍 Argentina