For this week's interview, I'm talking with Ankur Tyagi. He's a software engineer who's worked at multinational companies like Volvo, Barclays, and Accenture. He grew up in Pune, India and now lives in Gothenburg, Sweden. Ankur is a prolific contributor to freeCodeCamp's open source learning resources. He also runs DevTools Academy, where he blogs about emerging developer tools. He shares tips for: - How he uses AI tools to get more done as a dev but... - He thinks leveraging AI is a skill any dev can learn, and we shouldn't worry about fewer dev jobs. - How to run you own developer consultancy - How writing programming tutorials can help you become a better engineer Support for this podcast comes from a grant from Wix Studio. Wix Studio provides developers tools to rapidly build websites with everything out-of-the-box, then extend, replace, and break boundaries with code. Learn more at https://wixstudio.com. Support also comes from the 10,889 kind folks who support freeCodeCamp through a monthly donation. Join these kind folks and help our mission by going to https://www.freecodecamp.org/donate News items: → freeCodeCamp just published our first-ever chess course, taught by a software engineer on our team who has an international ELO rating of 2285, putting him among top competitive players. Ihechikara Abba will teach you how to think strategically and checkmate your opponents. This beginner-level course starts off with algebraic chess notation and identifying the squares. Then you'll learn several endgame patterns. We published both a handbook and an accompanying YouTube course for you to reference and share with your friends. (full length handbook): https://www.freecodecamp.org/news/checkmate-patterns-in-chess-for-beginners → freeCodeCamp also published this comprehensive course on how to build your own AI shopping agent. Software Engineer Ania Kubów will teach you how to use Node, TypeScript, LangChain's LangGraph, Gemini, MongoDB, and other popular tools to build your agent. By the end of this course, your agent will be able to autonomously perceive, plan, act, and respond to your users. It will also be able to decide when it has enough information to respond, and when it needs to first reach out for external information by searching product databases. (2 hour YouTube course): https://www.freecodecamp.org/news/building-an-ai-powered-e-commerce-chat-assistant-with-mongodb/ → Last month Alibaba dropped the latest version of their Qwen LLM and already the freeCodeCamp community has a comprehensive course on how to train it from scratch. You'll learn about its architecture, Training Hyperparameters, Muon Optimization, RoPE Positional Embeddings, inference, text generation, and more. (1 hour YouTube course): https://www.freecodecamp.org/news/code-and-train-qwen3-from-scratch/ → freeCodeCamp also published this guide to the field of System Design, written by a who applies the principles both to software development and to his day-to-day life. You'll learn about scalability issues, the CAP Theorem, Caching & CDNs, Rate Limiting, and other key concepts. (50 minute read): https://www.freecodecamp.org/news/learn-key-system-design-principles-behind-high-traffic-platforms-like-gaming-and-job-discovery/ → This week YouTube channel Gamer's Nexus posted a 3 hour documentary about China and the Graphics chips they've been using to train their AI models. The video has been copyright claimed multiple times by Bloomberg and taken down, but people keep re-uploading it. It's pretty awesome. The channel host travels to Hong Kong, Taiwan, Shenzhen, and Singapore to interview people along the process who are obtaining this hardware both legally and illegally. Note that this link I'm sharing may already been taken down by the time you click it, but you can search "gamers nexus GPU" and you should be able to find it. 3.5 hour watch: https://www.youtube.com/watch?v=kkNBZeS1OHk → The open source maintainer of the extremely popular Curl command line tool posted this excellent talk about AI slop security issues he keeps getting. He explains how people seem to be going into LLMs and asking "Find me a vulnerability in the Curl project. Then write a bug bounty submission for me and make it sound as alarming as possible." LLMs keep hallucinating new bugs, which lazy people then submit. Reviewing such bugs is now taking up more and more of his day. Since security is pretty important, he kind of just has to take these submissions seriously. A lot of non-native English speakers use LLMs to make their writing sound more native. But of course, most developers can tell when something's been written by an LLM. At the end of the talk, someone in the audience asks him point blank: has he ever received an issue that reads like LLM output that turned out to be a real security issue? And he says point blank: no. The lesson: don't use LLMs to re-write your issues or pull requests for you, or developers may assume the whole thing is just AI slop. 50 minute watch): https://www.youtube.com/watch?v=6n2eDcRjSsk This week's song of the week is 1991's Breather by UK band Chapterhouse. Absolute banger drums on this one. Link's in the description. Listen to it after the podcast. By the way if you're wondering, that's a sleeping cat on the cover: https://www.youtube.com/watch?v=ZGWaBJc1UrY
