Skill UI

A comprehensive guide and implementation examples for handling API rate limits (HTTP 429) when interacting with the Perplexity Sonar API. This resource provides multiple robust techniques, including exponential backoff with jitter, queue-based rate limiting, and custom Token Bucket implementation, ensuring reliable and high-throughput API calls for developers using TypeScript and Python.

This skill provides a comprehensive pattern for optimizing bulk API interactions. It enables efficient processing of large arrays of operations by supporting both synchronous (for small batches) and asynchronous (for large batches) execution. Key features include concurrency control, robust partial failure handling, progress tracking via background workers, and dedicated validation steps, ensuring reliable data throughput for high-volume requests.

Serve production-grade LLM APIs with vLLM, leveraging PagedAttention, continuous batching, quantization, and tensor parallelism to maximize throughput while keeping latency, GPU usage, and scalability under control for OpenAI-compatible deployments.

Optimizes large language model inference on NVIDIA GPUs with TensorRT, delivering 10-100× faster throughput, quantized precision (FP8/INT4), multi-GPU scaling, and serving-ready tooling for production deployments.

Automates throughput calculator tasks for Performance Testing by offering step-by-step guidance, industry best practices, and production-ready code for load, stress, benchmarking, and monitoring scenarios.

A comprehensive utility suite for handling external API rate limits, specifically designed for services like TwinMind. It implements advanced patterns including exponential backoff with jitter and request queue management. Use this to ensure resilient, high-throughput, and stable interactions with resource-constrained AI APIs, minimizing 429 errors.

This skill provides robust strategies and Python implementations for dealing with API rate limiting (HTTP 429 errors) when interacting with services like Vast.ai. It includes automatic retry logic, exponential backoff algorithms, and optimized techniques for batch operations and polling, ensuring stable and efficient API throughput.

A Python materials science library for constructing, analyzing, and converting crystal structures, computing phase diagrams and electronic properties, and integrating with the Materials Project API for high-throughput workflows.

This skill is designed for performing local, on-premise evaluations and benchmarking of models hosted on the Hugging Face Hub. It facilitates running specialized evaluation frameworks like `inspect-ai` and `lighteval` against local hardware. Users can select optimal inference backends, including `vllm` for high throughput, or fall back to Hugging Face Transformers or `accelerate`, ensuring comprehensive smoke testing and task selection for LLMs. It focuses purely on the local execution process, separate from remote jobs or publishing workflows.

Collection of battle-tested PyTorch training recipes covering architecture selection, optimizers, LR scheduling, precision tips, and debugging to train or fine-tune LLMs, vision, diffusion, and biomedical models with high GPU throughput.

This skill provides comprehensive composable handlers for Golang structured logging using the samber/slog ecosystem. It enables developers to build robust, multi-stage log pipelines by integrating sampling for throughput control, formatters for PII scrubbing, complex routers, and multiple backend sinks (e.g., Sentry, Loki, Datadog). It is essential for building high-observability Go applications.

This skill provides integration with the Adaptyv Bio Foundry API, enabling the automated design, submission, and analysis of protein experiments. Users can submit amino acid sequences for various high-throughput assays, including binding kinetics (BLI/SPR), thermostability, and expression yield. It guides the entire research workflow, from sequence input to final results retrieval, accelerating biopharma discovery.