Watch
Explainers, tutorials and research breakdowns from curated channels. Nothing loads from YouTube until you press play.
22:46A 22-minute AI Engineer talk for inference engineers that covers runtime tricks like paged KV cache and chunked prefill, then system-level disaggregation and kernel-writing agents.
AI Engineer · 1.2K views ·
14:22A 14-minute talk from the OpenHands CEO comparing four ways to sandbox coding agents, for developers who run agents with broad permissions.
AI Engineer · 3.9K views ·
19:17A 19-minute Fireworks AI talk on training open models on your own data and redeploying them in a continuous loop, for teams weighing closed-model costs.
AI Engineer · 6.1K views ·
14:07A 14-minute LangChain talk on finding, reproducing and fixing agent failures using production traces, for engineers who run agents in production.
LangChain · 8.1K views ·
43:05A 43-minute freeCodeCamp.org course that takes n8n from first workflow to agents, Slack approvals, error handling and cost monitoring, for builders new to the tool.
freeCodeCamp.org · 111K views ·
44:26A freeCodeCamp.org walkthrough for developers who want to read, train and RL-tune a small GLM-5.3-Flash-style model in Python, with all code on GitHub.
freeCodeCamp.org · 150K views ·
9:04A nine-minute IBM Technology explainer for engineers who need the basic vocabulary of distributed LLM inference: the four parallelism types and prefill versus decode.
IBM Technology · 107K views ·
20:12A 20-minute talk for engineers about to fine-tune a model to fix wrong answers: a clear rule for what belongs in the prompt, in retrieval and in weights.
AI Engineer · 1.9K views ·
20:32A 20-minute talk for engineers serving open LLMs: a layer-by-layer list of inference optimizations, from NVFP4 to speculative decoding, after a Nebius product intro.
AI Engineer · 2.6K views ·
17:04Ivan Leo of Google DeepMind explains the Gemini Interactions API and Managed Agents in 17 minutes, useful for developers building agents on Gemini.
AI Engineer · 8.1K views ·
20:10Jeremy Adams of Neo4j builds a wearable agent on a Raspberry Pi with NanoClaw, WhatsApp and graph memory, a 20-minute talk for builders of small personal agents.
AI Engineer · 8.6K views ·
18:01An 18-minute talk for engineers who run LLM apps in production: which cost, safety and quality metrics to track beyond latency and error rates.
AI Engineer · 7.7K views ·
11:35A 12-minute Docker talk for developers who let coding agents run unattended: why an agent's built-in guardrails fall short and how microVM sandboxes contain an agent.
AI Engineer · 5.9K views ·
16:49A 17-minute talk for anyone moving a browser agent from demo to production: the math of compounding step errors and an architecture that works around it.
AI Engineer · 4.2K views ·
12:23For developers whose coding or review agents misjudge new code: why a search tool alone falls short, and how Exa trims pages down to what the model needs.
AI Engineer · 7.5K views ·
16:23Sam Witteveen tries two open decision models, ImaJev 4B and Jev Omni, on forms and screenshots for RPA, useful for automation builders who need a model to make yes/no or multiple-choice calls.
Sam Witteveen · 30K views ·
56:14A 56-minute Weaviate Podcast episode for search and RAG engineers on using an LLM's attention over a million-token context as the retriever, and where that breaks.
Weaviate vector database · 489 views ·
17:11Sam Witteveen walks through Gemini 3.8 Flash TTS, from voice design and cloning to pricing, with Colab code, for developers choosing a text-to-speech model.
Sam Witteveen · 87K views ·
13:10Sam Witteveen explains speaker diarization and runs NVIDIA's Nemotron 3 Diarization on a full podcast, a 13-minute look for anyone building multi-speaker transcription.
Sam Witteveen · 52K views ·
11:08An 11-minute Hugging Face walkthrough of Transformers.js 4.3 for JavaScript developers who need models in the browser to return valid JSON every time.
Hugging Face · 4.5K views ·
28:22A 28-minute beginner's explanation of flow matching on the Hugging Face channel that leans on visuals over equations, for people who want the intuition behind modern image generators.
Hugging Face · 5.7K views ·
6:11A six-minute Hugging Face quickstart for Pi users who want a private, offline coding agent running on a local GGUF model through llama.cpp.
Hugging Face · 11K views ·
28:34A 28-minute walk through how Mem0 stores, ingests and retrieves agent memories, for developers who want long-term memory that goes past a single vector search.
Hugging Face · 60K views ·
17:16A 17-minute explainer on what prompt caching stores and the agent-code mistakes that break it, for developers paying for long AI agent sessions.
Hugging Face · 23K views ·