Skill UI

Guides engineers through DeepSpeed distributed training and NVMe I/O optimizations, covering ZeRO, pipeline parallelism, mixed precision, and DeepNVMe handles for synchronous and asynchronous tensor storage.

This skill acts as an expert guide for setting up comprehensive debugging environments, implementing distributed tracing solutions, and establishing robust diagnostic workflows. It is ideal for diagnosing complex, multi-service issues in development or production environments by standardizing logging, tracking correlation IDs, and ensuring end-to-end observability.

Orchestrates PyTorch-native TorchTitan 4D parallelism (FSDP2, TP, PP, CP) for large-scale LLM pretraining, scaling from 8 to 512+ GPUs with Float8, torch.compile, and distributed checkpointing to boot Llama 3.1, DeepSeek V3, or custom models.

This skill provides comprehensive guidance on implementing distributed tracing using industry standards like OpenTelemetry and tools such as Jaeger/Tempo. It enables developers to monitor the end-to-end lifecycle of a request across multiple microservices. Use this skill to debug latency issues, visualize complex service dependencies, pinpoint performance bottlenecks, and accurately trace error propagation in highly distributed systems.

Guides automated distributed training setups with step-by-step practices, production-ready configs, and validation checks for ML training workflows.

This skill provides expert guidance for building robust, scalable, and fault-tolerant applications using Elixir. It specializes in mastering OTP patterns, Phoenix LiveView, concurrent programming, and distributed system design. Use it for complex Elixir development tasks requiring best practices and high reliability.

This skill provides expert capabilities for diagnosing and resolving complex system failures in distributed environments. It is designed for investigating production incidents, performing deep root-cause analysis using observability tools (logging, tracing), and proposing robust, preventive fixes. Use it to significantly enhance system stability and reliability.

This skill is designed for advanced error detective work. It analyzes complex log streams, correlates stack traces across distributed systems, and detects anomalous error patterns. It helps identify root causes by correlating errors with system deployments and changes, providing actionable findings for immediate fixes and long-term prevention strategies.

This skill provides expert-level diagnostic capabilities for complex distributed systems. It specializes in analyzing production incidents, identifying deep root causes, and designing comprehensive observability solutions. Use this when investigating recurring errors, reviewing system logs/traces, and implementing preventative reliability measures.

Provides a comprehensive framework for implementing observability in Evernote integrations. It covers metrics collection (Prometheus), structured JSON logging, distributed tracing (OpenTelemetry), and defining health/readiness endpoints. Use this when setting up production monitoring, alerting, and performance tracking for mission-critical Evernote APIs.

This skill guides the design and implementation of complex GraphQL APIs. It specializes in schema design, utilizing Apollo Federation for distributed microservices, optimizing resolvers with DataLoader for N+1 prevention, and implementing real-time subscriptions. It covers security, complexity analysis, and adherence to best practices for building scalable, high-performance graph architectures.

HuggingFace Accelerate adds distributed support to any PyTorch script with just four lines, providing a unified API for DDP, DeepSpeed, FSDP, and Megatron, plus automatic device placement, mixed precision, and simple launch/config commands for fast prototyping.