103 links · newest first
A Grow with Google enrollment page for the Google AI Professional Certificate. The page promotes building AI fluency through 20+ hands-on activities, practicing real business challenges, and learning how to use AI as a partner in daily work. It also advertises three months of Google AI Pro for enro…
Visit originalAn XDA article about replacing cloud AI with a local large language model running in Proxmox. It describes why the author shifted to a self-hosted setup and how that change made the home lab more useful and satisfying. The page frames the setup as a practical way to keep AI workloads under the user…
Visit originalA homepage for LearnAI Feed, a daily-updated archive of AI learning links grouped by topic. The page highlights that the site organizes signal links for browsing and search, and includes navigation for a newsletter and sign-in. The visible content suggests a curated feed focused on making AI learni…
Visit originalThis article is a beginner-friendly guide to Google’s Gemini AI, focusing on practical ways to use it effectively. It presents 12+ copy-paste prompts and explains the parts of Gemini worth knowing, with examples tested across Gmail, Drive, Chrome, YouTube, Photos, and NotebookLM. The page is aimed …
Visit originalA Coin Bureau post about using Grok to analyze cryptocurrencies or stocks like a Wall Street analyst for free. It says the post shares six prompts intended to replace about $800 per year in research tools and encourages readers to save the post. The page is an X post from August 25, 2026.
Visit originalA video conversation with Greg Isenberg and Vas from Varick Agents about AI forward deployed engineering (FDE) and how to break into the role within 30 days. It explains why deployment, not model access, is the competitive advantage, tracing the FDE idea back to Palantir and describing the mix of t…
Visit originalA video by Nate B. Jones explaining why forward deployed engineer is one of the hottest AI jobs and what the role actually involves. It discusses how the major AI labs are hiring humans to help models work inside enterprise workflows, why pay bands vary so much, and which parts of the job come from…
Visit originalAdam Jacob explains what a software factory is and how to build one for both new and existing codebases. The talk covers agile manifesto principles, black-box testing for trust and safe refactoring, and what he calls the core secret behind effective software factories. It was shared via the Swamp C…
Visit originalThe video explains fine-tuning and demonstrates how to install and use the open-source Unsloth Studio locally and offline. It shows how to select models and datasets, fine-tune Qwen models with QLoRA, adjust training parameters, and monitor training runs. It also demonstrates creating custom questi…
Visit originalMihailo Zoin presents ten of 24 protocols for using Gemini Notebook (NotebookLM) more rigorously. The protocols address source organization, deep analysis, contradiction detection, decision testing, research paralysis, bias, and preventing AI-generated consensus from being mistaken for corroboratio…
Visit originalGoogle recommends seven NotebookLM prompts spanning student, business, and personal tasks, including generating quizzes, creating study guides, extracting syllabus URLs, analyzing revenue, researching competitors, organizing receipts, and modeling property purchases. The prompts demonstrate Noteboo…
Visit originalSpecification engineering expands prompt engineering by defining an AI task's objectives, inputs, constraints, outputs, edge cases, evaluation criteria, tests, and failure modes. The article explains its importance for coding agents, research assistants, data workflows, and structured outputs, citi…
Visit originalThis tutorial explains how to build a fully local command-line AI agent using Python, Ollama, and Alibaba's Qwen 2.5 model. It demonstrates defining a shell-command tool with Python's subprocess module, mapping it to Ollama's tool-calling schema, and maintaining a conversation loop that handles use…
Visit originalThe video argues that AI slop shifts the burden of reviewing unclear, unchecked writing onto whoever reads it next, costing hours every week. It explains that AI models converge toward broadly rewarded styles, so a universal anti-slop checklist cannot create distinctive authorship. It advocates an …
Visit originalThe article explains autoregressive transformer generation and the distinction between prefill, which processes the full prompt, and decode, which generates tokens one at a time. It demonstrates a simple key-value cache that avoids recomputing attention keys and values, reducing decode attention co…
Visit originalThe article explains how AirLLM runs the full 2.8-trillion-parameter Kimi K3 model using 3.72GB of VRAM through per-expert streaming and delayed MXFP4 decompression. It covers hardware, storage, disk-streaming limitations, setup requirements, installation commands, and Python code for generating te…
Visit originalThis Substack post promotes a free 367-page guide on LLM system design, described as a production-grade interview handbook for designing, serving, and scaling large language model systems. It says the guide includes 50 chapters across 13 parts, plus an offline companion project runnable on a person…
Visit originalThis X post promotes a 12-page PDF and full course on 'Graph Engineering' for multi-agent systems, attributed to a senior Anthropic engineer. It argues that an agent's memory dies with its context window, and that a knowledge graph makes memory permanent through an Extract, Resolve, Assemble, Query…
Visit originalThis X post highlights a free one-hour Stanford lecture on building LLM architectures from scratch, contrasting it with the $750,000+ salaries Anthropic reportedly pays engineers with this skill. It promotes the lecture as a must-watch resource and points to an accompanying article covering five co…
Visit originalThis X post explains five commonly misunderstood AI concepts: hallucination, context window, RAG, agents vs. chatbots, and fine-tuning vs. prompting vs. RAG. It clarifies what each term actually means, why people commonly misunderstand them, and offers practical usage tips such as verifying outputs…
Visit original