#semantic-kernel

Add a human reviewer to AI workflows: two integration patterns, implementation tips, trade-offs, and C# sketches.

Validation Agent: validate intermediate results, provide proofs-of-correctness, and gate actions in multi-agent orchestrations.

Agent-to-Agent (A2A) protocol: discovery, message schema, orchestration patterns, and security for orchestrating multiple agents

Ingestion → embedding → vector store → retrieval → response workflow for a Knowledge Agent with mail, OneDrive and SharePoint

Microsoft Semantic Kernel. Draft emails, propose meetings, and create calendar blockers.

How to orchestrate agents in Semantic Kernel: definitions, planning, and multi-step workflows.

Create Graph webhooks for shared mailboxes with app perms, validation, renewals, and secure processing.

Subscribe to Windows Event Log in a Windows Service; handle lifecycle, filtering, and SK processing.

Watch directories for changes; debounce events, recover with scans, and process files reliably with SK.

Implement scheduled jobs with Worker Service, cron/systemd, or Functions; avoid overlap and handle drift.

Handle webhooks safely: signature verification, idempotency, retry handling, queuing, and SK invocation.

Build HTTP-triggered SK endpoints with ASP.NET Core minimal APIs, auth, DI, resiliency, and hosting.

Overview of trigger types (HTTP, Webhook, Timer, FS watcher, Windows Event Log) and typical hosting options.

You don't need Cosmos DB to add memory. Postgres (pgvector) and Qdrant are great options — here's how in Semantic Kernel.

Designing user, team, and company memory with Azure Cosmos DB vectors, ACL filters, TTL, and C# snippets.

What I mean by memory: session, user, team, and company memories — and what developers must consider (permissions, privacy, retention).

Deep‑dive on SK plugins: structure, manifests, metadata, packaging, and governance best practices.

Deep‑dive on MCP functions: exposing capabilities via Microsoft Copilot Protocol, setup, security, and pros/cons.

Deep‑dive on SK OpenAPI functions: importing specs, auth wiring, errors/retries, and when to prefer native code.

Deep‑dive on SK prompt functions: design patterns, grounding, inputs, safety guardrails, and evaluation.

When and how to use native SK functions, .NET DI lifetimes, OBO auth, HttpClientFactory, and resiliency with Polly.

Overview of SK function types (Native, Prompt, OpenAPI, MCP, Plugins), when to use each, and .NET DI lifetime basics.

SQL Server vector storage and search SQL Server 2025 with DiskANN

Using Azure Cosmos DB as an integrated vector store (DiskANN) for chat sessions, memory, and retrieval

What Qdrant is, why it’s great for vector search, how to run it with Docker, and how to use it with Semantic Kernel in C#.

Getting Started with AI Search

How to chunk your data, filter it smartly, and combine BM25 with vectors for reliable retrieval.

Practical RAG patterns that work in production and how to evaluate them with recall, groundedness, and latency.

A practical primer on cosine similarity and common ANN indexes: HNSW, IVF, and PQ — with small, concrete examples.

Getting Started with Vector Search

Build a streaming chat endpoint with Server-Sent Events (SSE) in ASP.NET Core and wire it to Semantic Kernel for token-by-token output.

Build an SPFx chat webpart powered by Azure Functions and Semantic Kernel with streaming, Graph plugin calls, and Entra ID auth.

DeepSeek chat locally and via Azure AI Foundry; setup, SK integration, streaming, and EU/US data residency tips.

How to chat with localhost

Connect to OpenAI and Azure OpenAI, stream tokens, compare differences, use Managed Identity, and deploy models in Azure.

What Semantic Kernel is, how it works, which languages it supports, and where I'm using it.