Video summary

Don't waste 2026 learning the wrong tech skills (Meta Engineer's Take)

Main summary

Key takeaways

News and Commentary

Summary of Main Arguments and Key Points

  • Coding-focused “traditional” job strategy is becoming less reliable in 2026. The presenter argues that the usual path—grinding interviews (e.g., LeetCode) and targeting big-tech new grad roles—is not working well because the entry-level hiring market has deteriorated.

  • The entry-level software pipeline is tightening, especially for CS majors. He cites stats indicating:

    • Recent grads have the highest unemployment rates in CS, higher than the average across all majors.
    • Big tech new grad hiring is declining, including a drop from about 15% in 2019 toward lower levels by 2026.
    • Entry-level job postings are decreasing, and the number of employees with 1–3 years experience is down significantly since 2022.
    • Ongoing tech layoffs are part of the backdrop, including a noted 10% Meta layoff the prior week.

  • AI hiring is growing—but mostly for roles that aren’t truly “junior.” He highlights that job postings mentioning AI have risen sharply (much faster than overall postings). However, many AI roles still require 3+ years of experience, meaning beginners must position themselves differently.

  • AI layoffs don’t eliminate engineering needs—just coding time. He contends that the rationale “AI makes engineers unnecessary” is only partly true:

    • AI may reduce time spent writing code.
    • But major engineering work also involves understanding codebases, aligning with product/design, and producing/working through technical specs—areas AI can assist with but not fully replace.
    • He predicts a 6–12 month “correction” where leadership realizes they still need engineers and starts hiring/reshuffling accordingly.

  • “Vibe coding” can mislead new builders. He compares this to AI “plug-and-play” app builders: while it’s easy to generate something quickly, most projects never make it to production. He argues the hardest work is the last-mile integration into real systems—so engineers who can handle that will be in demand.

Recommended Plan for New Engineers (0–3 Years)

Phase 1 (0–6 months): Build AI-native projects instead of over-optimizing for interviews

  • He advises new grads to pause or deprioritize big-tech interviewing because it’s an uphill battle right now.
  • Emphasizes building real products/projects using AI, with LLMs as a core capability.
  • Suggests keeping LeetCode to ~10–15% of time (or even lower temporarily) since it’s historically not a strong reflection of real big-tech engineering work; it may still matter somewhat, but won’t dominate as much.
  • Encourages thinking beyond big tech: the shift enables more solo founders/startups, since building has become easier.

Project ideas aligned to near-future AI skills:

  1. RAG systems (retrieval over personal/company knowledge bases; embeddings + retrieval for workflows like research/notes/chatbots over internal data).
  2. Autonomous agents for multi-step workflows (example: an AI content research pipeline).
  3. Eval harnesses for agents (how to evaluate outputs, inspect behavior, and enable self-correction in production).

Phase 2 (~6–18 months): Target startups for experience

  • After building a portfolio, he recommends applying to startups, arguing they’re still hiring entry-level roles.
  • Reasons: faster learning, less siloing, and broader engineering exposure than large-company codebase immersion.
  • He frames startup work as a way to accelerate toward roles requiring 2–3 years of experience.

Phase 3 (12–18+ months): Re-enter the big-tech path as roles open

  • He expects more hiring to resume within this window due to engineering demand that AI doesn’t fully replace.
  • With startup/project experience, new grads can qualify for more conventional roles.

Practical “Do This Now” Steps

  • Reduce LeetCode to ~10–15% (or stop for a few months) until close to actually interviewing.
  • Pick one project you’d use yourself; build a quick version by end of week using AI, but learn actively rather than letting AI do everything.
  • Ship publicly: deploy it, publish code (GitHub), and ideally get a real URL/App Store presence if applicable.
  • Share for feedback on LinkedIn/X and other communities.
  • Join or build a peer community focused on the same journey.
  • Stop mass applying for ~3 months, arguing time is better spent building skills/projects during a weak hiring cycle.

Core Takeaway

  • His mantra: “build, ship, repeat.”
  • While the hiring market is broken for new grads, he argues the career strategy should pivot to AI-native building and portfolio creation, leveraging a potential hiring “backswing” later.

Presenters/Contributors

  • Jason (Meta Engineer; presenter and narrator)

Original video