Most important take away

Job hunting in tech is a numbers game and rejection is rarely personal — Maddy interviewed at 12 companies and got 4 offers after being rejected hundreds of times across her career. The biggest career risk of staying in a comfortable big tech role is building skills tied to internal stacks that don’t transfer, so engineers should periodically interview externally and keep their fundamentals (LeetCode, system design) sharp even when not actively looking.

Summary

Actionable insights and career advice:

• Treat job searching as a numbers game. Apply broadly and quickly. Maddy applied to 12 companies and landed 4 offers despite “hundreds” of lifetime rejections. Don’t self-reject — let the company tell you no.
• Apply the moment a req opens. Recruiters typically review only the first batch of applicants and toss the rest, so submit a “good enough” resume rather than waiting to perfect it.
• Use referrals strategically. Ask people who actually know you (college friends, ex-coworkers). Random referrals fail because referral forms ask how the referrer knows the candidate.
• Beware “golden handcuffs” and internal-only stacks. Working solely on proprietary tooling (e.g., Google’s internal frontend stack instead of React) makes your skills non-transferable. Periodically check that you could perform in interviews elsewhere.
• Manufacture a catalyst to actually leave. A four-hour corp-access bug pushed Maddy to start interviewing after 4+ years of vague “maybe I should look.” Schedule the search; don’t wait for anger.
• Interview prep stack:
  • LeetCode mix of easy/medium/hard. Don’t just read solutions — type them out and run them so concepts stick.
  • System design: Hello Interview YouTube channel (level-tagged: mid/senior/staff) and Grokking the System Design Interview.
  • Practice with peers running mock interviews — accountability beats solo studying.
  • For AI-allowed interviews (Meta is trialing), practice writing code with Llama/LLM assistance.
• Time management while employed: Time-block evenings on the calendar, use Q4 (Thanksgiving/Christmas slowdown) when work load is lighter, and treat the grind as temporary.
• Push for promotion before you feel ready. Maddy’s manager nominated her a cycle earlier than she’d have nominated herself. Ask peers who’ve been promoted to share their packets as a sanity check; do a gap analysis against the next-level rubric. Let the system reject you, don’t self-reject.
• Imposter syndrome reframe: You see your own failures plus everyone else’s curated successes — that’s an unfair comparison. Build a support network (especially women-in-engineering groups) and talk about failures openly.
• Non-traditional backgrounds work in big tech. School matters mainly for the first foot in the door; after that it’s irrelevant. Many of Maddy’s Google peers came from state schools or community college transfers.

Tech patterns and observations:

• Google’s tradeoffs: great perks, slow productionization due to layers of approval and heavy in-house tooling. Ads team moves faster than Search due to experimentation culture (5% traffic A/B tests). Google ships fast when leadership flags something a priority (Gemini after ChatGPT) — competition is the forcing function.
• AI workflows for productivity:
  • Claude Projects / ChatGPT Projects to structure long-form content from voice-dump brainstorms.
  • Voice dictation → LLM organization beats trying to write structured prose directly.
  • Claude Code and Cursor for fast side-project ideation.
  • LLM tone-shifting on emails (e.g., make apologetic drafts more assertive).
• Software engineering in 5 years: Jobs persist but shift toward prompting and tool orchestration vs. pure coding.
• AI risks worth more attention: prompt-injection vulnerabilities (Comet leaking bank info via UI prompts), AI-generated content indistinguishable from reality (deepfake voices for scams, election manipulation), and weak government technical literacy for regulation. Watermarking exists but is unreliable in the wild.
• Avoid “AI-washing”: Don’t shove AI into products where it doesn’t fit just to use the buzzword.

Chapter Summaries

• Why leave Google: Bureaucracy, an internal-only tech stack that wouldn’t transfer, and watching friends leave one by one. The realization that her skills wouldn’t survive a sudden layoff was a key driver.
• Golden handcuffs and Google culture: Free meals, gym, and on-campus life made her dependent on the employer. She wanted to “become an adult” with a life independent of the office.
• Engineering culture critique: Lots of config work on internal tools, slow productionization, but Ads team moved faster via experimentation. Google still competitive long-term when leadership prioritizes (e.g., Gemini response to ChatGPT).
• The catalyst to leave: A four-hour corp access bug during a 5K run triggered the realization she needed a backup plan. That moved her from passive consideration to active interviewing.
• The interview gauntlet: 12 companies, 4 offers. Funny incidents (interviewer’s model went down mid-interview; another interviewer’s office lost power). Reinforced that interview outcomes have heavy randomness.
• Tips for landing offers in a tough market: Apply early, use real referrals, embrace rejection as non-personal, grind LeetCode + system design (Hello Interview, Grokking the System Design Interview), do peer mock interviews.
• Managing time and energy while interviewing employed: Time-block evenings, exploit Q4 slowdown, lean on a friend study group for accountability, treat it as temporary.
• Imposter syndrome: Defined as feeling you’ve faked your way in. Antidote: recognize you compare your failures to others’ curated successes; find peer/mentor support; women-in-engineering groups help.
• Performance and promotion: Her manager pushed her up a cycle early. Strategy: ask peers for their promo packets, run a formal gap analysis against the next-level rubric, let the system reject you rather than self-rejecting.
• Non-traditional backgrounds in big tech: Many Google peers came from state schools or community college; school matters only for the first foot in the door.
• Content creation alongside engineering: Started by accident in the pandemic, found the women-in-SWE community, met her best friend via DMs. Plans to keep SWE as primary career — too risky to depend on an algorithm for income.
• AI takes: Stop bolting AI onto everything just for the buzzword. Real concerns: prompt-injection security holes (Comet bank info leak), undetectable deepfakes affecting elections and scams, weak governmental technical capacity for regulation. SWE in 5 years will be more about prompting and tool use than raw coding.
• AI workflows she uses: Claude Projects to structure videos from voice dumps, Claude Code/Cursor for side projects, LLM tone-shifting for emails. Notes models are still too sycophantic.
• Closing: If not a SWE, she’d be a marine biologist (won the Ocean Science Bowl nationals in high school).