QYTANG EVOLUTION
Explore complete public sessions by year and follow how technology evolves.
This episode introduces Google A2UI, where agents produce declarative JSON interface descriptions instead of executable HTML or JavaScript. It covers…
This episode introduces Nixtla TimeGPT for time-series forecasting, covering API tokens, data format, forecast calls, horizons, and visualization. It…
This episode explains Agent Skills in the context of LangChain Deep Agents. It packages domain instructions, metadata, scripts, and templates as reus…
This episode evaluates AIHubMix, Alibaba Cloud Bailian, and SiliconFlow through practical NetDevOps tasks. It covers API integration, basic text gene…
This episode builds an executable operations agent with OpenClaw. It connects task understanding, tool use, skill loading, and execution feedback int…
This episode introduces Hugging Face Smolagents and its code-agent approach. Learners see how an agent decomposes a goal, writes and executes code, c…
This episode builds a task-specific custom GPT through instruction design, knowledge files, capabilities, conversation starters, and iterative testin…
This episode maps the roles of Skills, subagents, commands, and hooks in Cursor. It shows how to package repeatable work, delegate bounded tasks, tri…
This episode walks through RAGFlow from deployment to application: account and model configuration, API keys, knowledge-base creation, document inges…
This episode introduces the OpenAI Agents SDK through agents, tools, handoffs, context, and the run loop. It demonstrates combining model reasoning w…
This episode introduces AWS Bedrock's managed foundation-model platform and unified APIs across Claude, Llama, and Mistral. It covers permissions, mo…
This five-part episode analyzes NVIDIA SuperPOD reference architectures for B200 and GB200 NVL72. It covers the system overview, key compute and inte…
This episode explains the Model Context Protocol host-client-server architecture and its JSON-RPC 2.0 resources, tools, and prompts. It compares stdi…
This episode converts private documents into embeddings stored in Qdrant and retrieves relevant context for GPT. It covers text chunking, embeddings,…
This two-part episode builds a question-answering bot with Hermes Agent and the DingTalk CLI. It covers agent and model configuration, tools and skil…
This three-part episode examines the bandwidth, latency, nonblocking-fabric, and congestion-control demands of AI/ML workloads; introduces specialize…
This episode interprets Cisco's AI Readiness Index across strategy, infrastructure, data, talent, and governance, then connects those findings to AIO…
This episode accelerates large-language-model inference with NVIDIA TensorRT-LLM. It covers model conversion, precision and quantization choices, opt…
This episode introduces NVIDIA NIM (NVIDIA Inference Microservices), which packages language, vision, and speech models as standardized deployable in…
This episode explores KAG knowledge modeling and retrieval augmentation through containerized deployment, reverse proxy and TLS, model and knowledge-…
This episode runs Llama 3 with Ollama and integrates it through LangChain ChatOllama, covering model download, service startup, parameters, streaming…
This episode explains Agent2Agent agent cards, capability discovery, task and message models, long-running jobs, and cross-framework interoperability…
This episode uses LangChain to extract structured data from network logs and other unstructured text through schemas, field descriptions, constrained…
This four-part episode reviews AI tools for data analysis with Noteable, slide creation with Gamma, video translation with Rask, and image generation…
This episode combines Drain3 online log-template mining with LogAI analytics to turn large volumes of unstructured logs into stable templates and com…
This episode calls cloud image-recognition APIs from Python, covering credentials, tokens, request encoding, image upload, JSON parsing, reusable wra…
This episode develops a multimodal RAG approach for images and text, focusing on Cohere Embed 4's unified embedding space, long context, Matryoshka d…
This episode deploys AnythingLLM with local knowledge bases, document ingestion, vector retrieval, model selection, workspace isolation, and multi-ag…
This episode discusses how AI changes networking, systems, and infrastructure roles through technical assistance, configuration automation, and new s…
This episode presents NetClaw as a network-operations agent that combines language models, network knowledge, tools, and controlled execution. It cov…
This episode introduces Google Agent Development Kit concepts including agents, tools, sessions, runners, and multi-agent composition. Python example…
This episode combines AI agents, Terraform infrastructure as code, and CI/CD into a human-in-the-loop change pipeline. It follows the process from re…
This two-part episode builds an AI foundation for network engineers across machine learning, deep learning, generative AI, and learning paradigms, th…
This episode accelerates PCAP and DNS analysis with GPU-CUDA. It covers parsing, feature preparation, batched parallel computation, and result aggreg…
This episode covers Neo4j nodes, relationships, properties, and Cypher for network topology, paths, and device modeling, then connects graph database…
This episode explains GPT function calling: declaring external tools with structured schemas, letting the model select tools and arguments, executing…
This episode uses Supabase PostgreSQL, authentication, storage, realtime features, and APIs as the backend of an AIOps platform, then connects AI ana…
This episode wraps Cisco Catalyst Center and Meraki management APIs as MCP tools so agents can query devices, inspect state, and invoke controlled op…
This episode explains Mem0's user, session, and agent memory layers, including fact and preference extraction, hybrid vector/key-value/graph storage,…
This episode combines GPT function calling with network data sources such as NetBox so natural-language questions invoke deterministic query function…
This episode uses Python for speech recognition, covering audio input, speech-to-text conversion, and programmatic invocation. Companion code integra…
This episode introduces Meta Llama Stack's standard APIs for inference, safety, memory, agents, evaluation, post-training, synthetic data, and provid…
This episode compares fixed-size, recursive, document-specific, and semantic chunking, using sentence embeddings and breakpoints to preserve meaning …
This episode introduces LangChain's models, prompts, chains, tools, and agents, connecting them with Zapier, GPT-4 plugins, and workflow platforms. I…
This episode uses OpenAI Swarm agents, function tools, context variables, and handoffs to build a lightweight multi-agent AIOps workflow for fault an…
This episode builds a closed-loop AIOps workflow where metric alerts and logs trigger context collection, AI diagnosis, and controlled automation too…
This episode combines GPT-4 function calling with NetBox, PyATS, and Elastic logs to build a Cisco network-automation assistant. Device, interface, O…
This episode compares two newer RAG approaches. HiRAG reduces noise and hallucination through hierarchical retrieval, selection, composition, and rea…
This episode covers Dify model management, prompt applications, workflow orchestration, knowledge-base RAG, tool calls, and publishing. It combines m…
This episode introduces LangChain Deep Agents for complex, long-running work through planning and TODOs, subagents, filesystem tools, context managem…
