Maddy Zhang

AI was supposed to replace software engineers. Instead, companies are discovering that AI might be far more expensive than they expected.

Billions are being poured into AI. Coding agents are burning through tokens. Companies are cutting back on the very AI tools engineers were told to use. And the AI bubble is starting to face a much bigger question: does the economics actually work?

In this video, I break down what’s happening behind the headlines from Uber burning through its AI budget to Microsoft changing its internal AI strategy and what it means for developers, AI jobs, and the future of software engineering.

🚀 What you'll learn:
✅ Why companies are spending far more on AI than expected
✅ What happened when Uber rolled out Claude Code to thousands of engineers
✅ Why Microsoft changed its internal AI coding tools
✅ Why cheaper tokens can actually create bigger AI bills
✅ How AI agents consume dramatically more tokens than normal chat
✅ Whether AI is actually cheaper than a human engineer
✅ Why AI productivity gains may have been oversold
✅ How AI is changing software engineering hiring
✅ Why token and compute budgets could become part of compensation
✅ What context engineering is and why it matters
✅ How model routing, caching, and better context management reduce costs
✅ Why knowing when not to use an AI agent may become a critical skill

👉 Watch the full breakdown here: https://youtu.be/2TgYw9wXv5s

4 days ago (edited) | [YT] | 485

Maddy Zhang

The tech job market isn't dead in 2026 but it has fundamentally changed.

While headlines focus on layoffs, hiring freezes, and AI replacing software engineers, the reality is much more nuanced. Some tech roles are becoming increasingly automated, while others are seeing record demand, higher salaries, and a growing shortage of qualified talent.

In this video, I break down the tech careers that are actually worth pursuing in 2026 based on hiring trends, salary data, AI adoption, and long-term career resilience. You'll learn which roles are growing, which ones I'd be cautious about, and how to build a career that stays valuable in the age of AI.

🚀 What you'll learn:
✅ Why the tech job market is splitting instead of shrinking
✅ Which tech careers are growing despite mass layoffs
✅ Why AI Engineer is still one of the highest-paying roles and how to stand out
✅ The difference between real AI engineering and simply calling AI APIs
✅ Why Data Engineering is one of the most underrated careers in tech
✅ Why Cybersecurity demand continues to grow in the AI era
✅ How AI Product Managers combine technical knowledge with business strategy
✅ Why Cloud & Platform Engineering remains critical as AI infrastructure expands
✅ The skills employers actually value in 2026
✅ Which tech jobs are becoming increasingly vulnerable to AI automation
✅ How to future-proof your career by focusing on judgment, ownership, and system design instead of repetitive work

👉 Watch the full breakdown here: https://youtu.be/aGnWRt6u-fg

1 week ago | [YT] | 674

Maddy Zhang

Went to Open Sauce (‪@williamosman‬’s science and engineering festival) earlier this month!

It’s always one of my favorite events of the year :) It’s an amazing mashup of makers, creators and nerds.

Some highlights:
-life sized Gameboy
-taser knife fighting
-the cutest R2D2
-soldering a Bop-it style game on my badge
-watching the World Cup finals at the ‪@youtubecreators‬ lounge
-hanging with my friends <3

Can’t wait to be back next year!

Skip the Carousel

1 2 3 4 5 6 7 8 9 10

2 weeks ago | [YT] | 663

Maddy Zhang

If you want to learn coding in 2026 but don't know where to start, this is for you.

The internet makes learning to code feel far more complicated than it actually is. One person says learn Python. Another says JavaScript. Someone else says AI will replace programmers before you even get a job.

The reality is much simpler.

In this video, I share the exact roadmap I'd follow if I were starting from zero today. You'll learn what to study first, what to ignore, how to avoid tutorial hell, how to build projects that actually matter, and how to use AI tools without becoming dependent on them.

Here's what I cover 👇
✅ Whether learning to code is still worth it in the age of AI
✅ Why Python is the best programming language to learn in 2026
✅ The complete beginner roadmap from zero to building real projects
✅ The programming fundamentals every developer needs to master
✅ How real software is structured beyond beginner tutorials
✅ The projects that actually help you become a better engineer
✅ The professional skills (Git, debugging, reading code) most courses skip
✅ How to use AI tools like ChatGPT and Claude the right way while learning
✅ The software engineering, AI, and data careers Python can unlock

👉 Watch the full breakdown here: https://youtu.be/aGnWRt6u-fg

2 weeks ago | [YT] | 944

Maddy Zhang

RAG (Retrieval-Augmented Generation) is one of the most important concepts in AI engineering but building a production-ready RAG system is much more than embedding documents into a vector database.

In this video, I explain how production RAG systems actually work, why basic RAG pipelines fail, and the techniques AI engineers use to build accurate, scalable, and reliable retrieval systems.

💡 Here's what you'll learn:
✅ What RAG (Retrieval-Augmented Generation) is and why it's used
✅ How the RAG pipeline works: Indexing → Retrieval → Generation
✅ Embeddings, vector databases, and semantic search explained
✅ Why naive RAG systems fail in production
✅ How chunking strategy affects retrieval quality
✅ Hybrid Search (Vector Search + BM25) for better accuracy
✅ Reranking to improve retrieval relevance
✅ Query Transformation and HyDE for smarter search
✅ Agentic RAG and how it enables multi-step reasoning
✅ When to use RAG instead of fine-tuning
✅ Best practices for building production-ready AI applications

👉 Watch the full breakdown here: https://youtu.be/9BpXVodhALk

3 weeks ago | [YT] | 449

Maddy Zhang

Before you spend money on another random AI course, learn directly from the company building some of the most widely used AI tools.

Anthropic, the company behind Claude, offers free courses covering Claude, Claude Code, APIs, MCP, agent architectures, RAG, tool use, and building reliable AI applications.

In this video, I break down the 6 best free Anthropic courses for developers and anyone trying to build practical, production-ready AI skills.

Here’s what I cover 👇
✅ Claude 101 and the features most users overlook
✅ How to use Claude Code effectively on real codebases
✅ Building AI applications with the Claude API
✅ Prompt evaluation, tool use, RAG, and agent architectures
✅ How to build MCP servers from scratch
✅ Advanced MCP concepts such as transports, sampling, and security
✅ Understanding AI capabilities, limitations, and context windows
✅ The best order to take these courses based on your experience

👉 Watch the full breakdown here: https://youtu.be/9Iw-UtQQe8c

1 month ago | [YT] | 917

Maddy Zhang

Most developers think Redis is just a cache but that barely scratches the surface of what it can do.

In real production systems, Redis is used to store user sessions, build live leaderboards, limit API requests, coordinate distributed jobs, power real-time messaging, and protect databases during sudden traffic spikes.

In this video, I break down how Redis works, why its single-threaded architecture makes it surprisingly fast, the most important patterns developers build with it, and the trade-offs you need to understand for system design interviews and real-world engineering.

💡 Here’s what you’ll learn:
✅ Redis fundamentals — why it is a data structure server, not merely a cache
✅ In-memory performance — how Redis achieves sub-millisecond response times
✅ Single-threaded architecture — why sequential command execution can improve speed and atomicity
✅ Caching — how cache-aside, TTLs, eviction policies, and cache stampedes work
✅ Rate limiting — how atomic counters control API request volume
✅ Leaderboards — how sorted sets maintain live rankings efficiently
✅ Session storage — how shared sessions work across multiple application servers
✅ Distributed locks — how Redis coordinates workers and prevents duplicate jobs
✅ Real-time messaging — when to use Pub/Sub versus Redis Streams
✅ Persistence — the differences between RDB snapshots and AOF
✅ Scaling — how replicas, sharding, Redis Cluster, and hot keys affect architecture

👉 Watch the full breakdown here: https://youtu.be/zrnRg4gu7cw

1 month ago | [YT] | 660

Maddy Zhang

Everyone keeps saying software engineering is dead in 2026. The job market is brutal, AI is changing entry-level work, and the old junior developer roadmap does not work the same way anymore.

But becoming a software engineer is still possible, you just need a different strategy.

In this video, I break down the five-stage roadmap I’d follow to become job-ready as a software engineer in 2026, from learning fundamentals and using AI properly to building real projects, preparing for interviews, and getting in front of hiring managers.

💡 Here’s what you’ll learn:
✅ Why the software engineering market has not disappeared — it has split
✅ Why the old junior developer roadmap no longer works
✅ What companies expect from new engineers in 2026
✅ Why you need to become useful faster than before
✅ Which programming language to choose first
✅ Why foundations still matter even in the AI era
✅ How to avoid getting stuck in tutorial hell
✅ How to use AI coding tools without becoming dependent on them
✅ Why AI-generated portfolios are not enough anymore
✅ What kind of projects actually prove your ability
✅ Why LeetCode is not dead yet
✅ How to prepare for system design as a beginner
✅ Why networking, referrals, and timing matter more than mass applying
✅ The exact five-stage roadmap I’d follow from zero

👉 Watch the full breakdown here: https://youtu.be/Egr6BiXpHQc

1 month ago | [YT] | 704

Maddy Zhang

Most developers using AI to code are doing it wrong not because they chose the wrong tool or because their prompts aren’t clever enough, but because they skipped the fundamentals that determine whether AI produces maintainable software or just creates more code to debug.

In this video, I break down the five principles that make AI coding genuinely useful. Whether you use Claude Code, Cursor, GitHub Copilot, Gemini, or another coding assistant, these principles will help you write cleaner, safer, and more maintainable code while keeping your engineering judgment firmly in the loop.

💡 Here’s what you’ll learn:
✅ Why clear thinking matters more than clever prompting
✅ How vague requirements cause AI to confidently build the wrong solution
✅ How to create persistent project context using rules and memory files
✅ What to include in files like CLAUDE.md to prevent repeated architectural mistakes
✅ Why large AI-generated changes create a compounding error surface
✅ How shrinking the blast radius makes AI-generated code easier to verify
✅ Why every AI coding workflow needs an objective feedback loop
✅ How test-driven development helps catch subtle AI-generated bugs
✅ Which coding tasks are safe to delegate to AI
✅ Why core business logic, data models, API contracts, and security decisions require human judgment
✅ How to use AI as an executor without becoming dependent on code you can’t explain

👉 Watch the full breakdown here: https://youtu.be/wQ46lax_ya0

1 month ago | [YT] | 485

Maddy Zhang

Most people think they’re good at AI because they use ChatGPT or Claude every day. But using AI frequently doesn’t mean you’re using it effectively and in some cases, adding AI to a bad workflow can actually make you slower.

In this video, I break down the 5 levels of AI proficiency every developer should understand in 2026. From using basic chatbots to building production-ready AI systems and orchestrating autonomous agents.

💡 Here’s what you’ll learn:
✅ Level 1: Consumer — using AI chatbots for everyday tasks, learning, writing, and debugging
✅ Level 2: Practitioner — building reusable prompts, workflows, projects, and custom instructions
✅ Level 3: Builder — creating real AI features with APIs, RAG, vector databases, and evals
✅ Level 4: Architect — designing reliable AI systems with routing, caching, fallbacks, and observability
✅ Level 5: Orchestrator — managing autonomous agents, specifications, testing loops, and AI workflows
✅ Why using AI more doesn’t always make developers faster or more productive
✅ How to move from basic prompting to building production-ready AI products
✅ Which AI level early-career and mid-level engineers should aim for
✅ Why judgment, systems thinking, and clear instructions matter more than tools

👉 Watch the full breakdown here: https://youtu.be/YvCQUx7LZ-o

1 month ago | [YT] | 879