Skill UI

A comprehensive guide and set of tools for establishing a local development workflow tailored for CoreWeave GPU deployments. This includes steps for building Docker containers, validating Kubernetes YAML manifests using dry-run, and managing the full deployment lifecycle for AI inference services before pushing to the cloud.

This skill outlines how to implement and manage Role-Based Access Control (RBAC) and namespace isolation for multi-team GPU workloads running on CoreWeave's Kubernetes platform. It provides guidance on setting strict resource quotas, preventing noisy-neighbor issues, and defining granular roles (e.g., ML Engineer, Inference Operator). This is essential for maintaining resource isolation, ensuring compliance (SOC 2, HIPAA), and managing access across large-scale cloud deployments.

This skill provides comprehensive observability for CoreWeave's GPU-intensive workloads running on Kubernetes. It monitors critical operational metrics, including GPU utilization, memory usage, inference latency (p99), and job completion rates. Use it for setting up DCGM metrics dashboards, configuring proactive alerts for underutilization or Out-Of-Memory (OOM) conditions, and ensuring high reliability for AI/ML deployments.

This guide provides expert strategies for optimizing GPU inference performance on CoreWeave infrastructure. It covers GPU selection based on workload (LLM, image generation, training), advanced techniques like continuous batching (using vLLM), autoscaling setup (HPA), and benchmarking data. Use this to maximize GPU utilization, minimize latency, and optimize large-scale AI model deployment.

A comprehensive checklist for ensuring GPU workloads, such as inference services or model training pipelines, are fully prepared for production deployment on CoreWeave. It covers critical MLOps and DevOps aspects including autoscaling, resource limits, security policies, persistent storage validation, monitoring setup, and rollback procedures.

This skill provides methods for managing GPU quota limits and rate restrictions for CoreWeave cloud services. It includes using `kubectl` to check resource quotas and demonstrates implementing robust request queuing mechanisms using Python's `asyncio` library, ensuring stable and efficient resource allocation for high-concurrency inference endpoints.

A comprehensive guide to advanced, production-ready LLM inference using LangChain 1.0. This skill addresses common pitfalls and bugs when integrating multiple providers (Claude, GPT-4o, Gemini), covering issues like structured output reliability, correct streaming token accounting, and robust handling of complex AIMessage content blocks. It teaches best practices for model initialization, provider routing, and version-safe code implementation.

A comprehensive reference guide for common errors encountered when interacting with the Together AI API. This includes error codes and solutions for inference, fine-tuning, and model deployment, covering authentication, rate limits, validation failures, and billing issues. Essential for developers integrating OpenAI-compatible LLM services.

A comprehensive workflow guide for fine-tuning open-source Large Language Models (LLMs) using the Together AI API. This process covers preparing structured training data (JSONL format), uploading the file, initiating the fine-tuning job, monitoring the training status, and finally deploying the custom model for inference. Ideal for developers needing to customize models for specific enterprise tasks, such as customer support or specialized content generation.

This guide covers essential security best practices for utilizing Together AI's OpenAI-compatible API for inference, fine-tuning, and model deployment. It provides code examples and architectural advice on secure API key management, robust webhook signature verification, input validation using schemas (like Zod), and implementing data redaction pipelines to protect sensitive information in logs and datasets. Essential for building secure AI applications.

Build and operate Python Azure IoT Edge modules for telemetry processing, local inference, and protocol translation. Includes deployment, observability, and production readiness checks.

A critical discipline for managing expenses when integrating paid AI or inference APIs. This guard treats dollar cost as a third complexity dimension, mandating the definition of cost contracts (max calls per run, max budget per run, max budget per day). It enforces best practices for bounding loops, retries, and fan-out, ensuring that code bugs cannot lead to unexpected overspending. Best practice requires configuring hard caps both in the code and at the provider dashboard for defense in depth.