Skill UI

Phoenix is an open-source observability platform for tracing, evaluating, and monitoring LLM applications; use it to debug prompts, run dataset evaluations, build experiment pipelines, and watch production inference in real time.

Fully managed, auto-scaling vector database tailored for production AI workloads such as RAG, recommendations, and semantic search, offering hybrid dense+sparse search, namespaces, metadata filters, and <100ms latency on serverless infrastructure.

Prediction Monitor Assistant automates machine learning deployment monitoring tasks, offering step-by-step guidance, best-practice patterns, production-ready code and configuration generation, and validation to keep serving and inference workflows reliable.

Document describes structured prompts, chain-of-thought guidance, few-shot examples, system prompt design, and tooling advice for building reliable prompt engineering workflows in LLM-augmented products.

Automates PyTorch training workflows by guiding data preparation, model configuration, hyperparameter tuning, and experiment tracking while following industry best practices for production readiness.

Qdrant is a high-performance, Rust-powered vector database designed for production-grade RAG (Retrieval-Augmented Generation) and semantic search applications. It provides fast nearest neighbor search, supports complex hybrid search using metadata filtering, and offers horizontal scalability through sharding and replication. Ideal for building real-time recommendation systems and enterprise-level knowledge retrieval.

Designs and implements production-grade RAG systems with document chunking, embeddings, vector stores, hybrid search, reranking, and evaluation, ideal for knowledge-grounded AI apps that need semantic retrieval, similarity search, and context augmentation.

Auto-activating assistant that guides AWS Sagemaker endpoint deployment, from best practices to production-ready code, configurations, and validation for inference serving and monitoring workflows.

Comprehensive skill covering the full lifecycle of visual AI systems. Expertise includes object detection (YOLO, Faster R-CNN), instance and semantic segmentation (SAM, Mask R-CNN), and video analysis. The workflow guides users through dataset preparation, selecting optimal architectures, training using frameworks like PyTorch, and achieving production readiness via model optimization (ONNX, TensorRT) and deployment on various edge or cloud targets.

Sentence Transformers provides 5000+ pre-trained models that generate sentence, text, and image embeddings for RAG, semantic search, clustering, and multilingual production workloads without relying on external APIs.

SGLang is a high-performance serving framework for LLMs and VLMs that uses RadixAttention prefix caching to deliver structured outputs (JSON/regex/grammar) and agentic workflows with function calls, providing up to 5× faster inference than vLLM across multi-GPU production deployments.

Guides post-training of GLM-class LLMs with Megatron-LM and SGLang, focusing on GRPO rollout generation, custom data buffers, and tight RL scaling workflows for research and production use.