Devops Engineer

Location : Remote  (TX)

Long term Contract (Full Time)

• This Engineer will play a key role in building, operating, and scaling a production-grade AI inferencing platform that supports high-throughput, low-latency large language model (LLM) workloads.
• The position combines software engineering, platform engineering, and reliability engineering responsibilities, with a focus on developing Python-based services, automating model deployments, and optimizing model serving performance using vLLM, Kubernetes, KServe, and Knative.
• The engineer will be responsible for containerization with Docker and Podman, maintaining and tuning PostgreSQL data models, troubleshooting production issues, and driving platform reliability through observability and automation.
• Working closely with ML engineers and platform architects, this individual will help scale mission-critical AI services, support new model rollouts, and continuously improve the infrastructure that powers enterprise AI applications. 

REQUIRED SKILLS AND EXPERIENCE

• Strong hands-on python coding & development experience as well as experience working with FastAPI

• Expertise with Docker & Kubernetes, including deployments, services, autoscaling, and

• troubleshooting workloads in production.

• Practical experience with Postgres, including query optimization and schema design.

• Cloud platform experience, such as Azure/AKS, AWS/EKS, or Google Cloud Platform/GKE.

• Strong debugging skills across the stack — from application code to infrastructure

• NICE TO HAVE SKILLS AND EXPERIENCE

• Direct experience with vLLM or LLM inferencing

• Experience with GPU-aware scheduling and resource management in Kubernetes.

• Familiarity with LLM-specific optimizations: speculative decoding, quantization (FP8/INT8), KV cache

• management, and continuous batching.

• Experience with API gateways or LLM routing layers, such as LiteLLM or similar.

• Background in SRE practices: SLOs/SLAs, incident response, and on-call tooling.

• Experience with Helm, ArgoCD, or other GitOps deployment tooling.