Validation Framework – Development, Maintenance, Testing (BB48FT RM 3802)

Position: Validation Framework – Development, Maintenance, Testing (BB48FT RM 3802)

Essential duties and responsibilities:

Design and implement efficient AI inference pipelines for production environments

Optimize model serving architecture for low-latency, high-throughput inference

Develop and maintain inference operations infrastructure and monitoring systems

Establish comprehensive metrics and monitoring for AI model performance

Develop frameworks for measuring latency, throughput, accuracy, and resource utilization

Conduct performance profiling and bottleneck analysis

Build reusable Python modules and frameworks for ML operations

Develop C wrapper libraries for performance-critical components

Create APIs and SDKs for model deployment and inference

Containerize ML models and services using Docker

Design multi-stage Docker builds for optimized container images

Implement orchestration solutions (Kubernetes, Docker Compose)

Manage container registries and deployment pipelines

Integrate vector databases with AI applications (RAG, semantic search, recommendation systems)

Qualifications:

Deep understanding of machine learning concepts, model architectures, and inference optimization

Experience with MLflow, Kubeflow, or similar ML platform tools

Hands-on experience with vector databases (Pinecone, Weaviate, Milvus, Qdrant, ChromaDB, or similar)

Strong proficiency in Python; experience with C/C++ for wrapper development

Proficient in Docker, container optimization, and orchestration platforms

PREFERRED:

Experience with LLM frameworks (LangChain, LlamaIndex, Hugging Face)

Experience with streaming inference and real-time systems

Knowledge of PCIe Gen4/5/6 technology is an advantage

Previous experience with storage systems, protocols, and NAND flash is an advantage

SKILLS:

Excellent interpersonal skills

Strong can-do attitude

*******************************************************************************************************************************************