Skill UI

A comprehensive guide to optimizing Groq API usage and managing inference costs. Learn techniques such as smart model routing (selecting smaller models for simple tasks), minimizing prompt tokens, batching requests to reduce overhead, and implementing caching for deterministic calls. This guide helps developers achieve significant cost savings and improve system efficiency when handling high-volume AI workloads.

Comprehensive guide for deploying applications powered by the Groq inference API. Covers best practices and specific setup instructions for multiple production environments, including Vercel Edge Functions, Google Cloud Run, Docker containers, and other containerized platforms. Essential for setting up robust, low-latency, production-ready AI services.

Guides production-grade miles RL training for massive MoE models, covering FP8/INT4 low precision, train-inference alignment, and speculative rollout workflows to maximize throughput in enterprise deployments.

A comprehensive guide to implementing advanced asynchronous patterns with Mistral AI. This skill covers using the Agents API for stateful, multi-turn workflows, the Batch API for cost-effective bulk inference, and event-driven streaming architectures. It is essential for building robust, long-running, or high-volume AI applications that require reliable background job handling.

This skill automates the setup and configuration of Ollama, allowing users to deploy Large Language Models (LLMs) locally. It performs system assessment (OS, GPU, RAM), selects optimal models based on hardware availability, and handles installation across macOS, Linux, and Docker. Use this when you require free, self-hosted AI alternatives, need offline inference capabilities, or wish to eliminate reliance on costly hosted API services.

Guides creation of persona-targeted VS Code CodeTour .tour files, covering repo exploration, intent inference, step composition, and validation so each narrative walkthrough is anchored to real files and tailored depth.

Modal is a serverless cloud platform for running Python code on demand, including GPU-accelerated AI/ML workloads. Use for deploying models, serving web APIs, batch processing, and scaling Python applications to cloud containers with automatic scaling from zero to thousands of instances. Supports custom container images, persistent storage, and sub-second cold starts for low-latency inference.

Automate the entire deployment lifecycle for ML inference services on CoreWeave GPU cloud. This workflow integrates unit testing, container image building (to GHCR), and rolling deployment to CoreWeave Kubernetes clusters via GitHub Actions. Ideal for ensuring stable, high-performance deployment of GPU-dependent models. Key features include validating GPU manifest requests and managing deployments using standard Kubernetes APIs.

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.