Navdeep Singh

3 Years ago I joined Amazon after graduation. It was not a great experience, so I left after two months.

It’s not like me to make irrational decisions like this, but for some reason every fiber in my body was telling me to quit.

Afterwards, I felt like I let everyone down. My mom had raised me and my two brothers by herself. And this was the job I worked my whole to get to, so that I could repay her. But now I had just thrown it away.

The only silver lining I had was I felt that I was reasonably smart and reasonably hard working.

If I got there once, surely I could do it again, right?

After a 5 month stretch of struggling with drugs and alcohol, I decided I would start making youtube videos.

I liked solving leetcode coding problems, so I started explaining my solutions on Youtube. I did this every single day and it felt gratifying to help others.

The job market was rough at the time, so I didn’t get any interviews for a whole year (seriously).

The one company who did respond to me, believe it or not, was Google. And I was fortunate enough to pass!

8 months ago I decided to leave Google so I could make videos and work on my own platform full-time. I loved Google so it was a tough decision, but it’s definitely not the worst decision I’ve ever made :)

Thanks for taking a chance on me Google, and thanks to all the incredible folks I met along the way!


#neetcode #leetcode #codinginterview

I first interviewed at google in 2019 and I failed. I’m not too embarrassed to share the mistakes I made because learning from this experience is the reason I was able to pass and join Google in 2021.

Here’s how each round went.


Round 1 - I misinterpreted the question and solved a slight variation of it. I only caught this after I had fully implemented the solution. While I was able to correct my solution towards the end, I think this still gave a bad signal.


Round 2 - I was able to implement the solution quickly, but my the interviewer mentioned the solution was more complicated than it needed to be. For some reason I assumed they wanted me to do the more complicated variation. In hindsight, I should've just asked the interviewer if the simpler approach was okay. It was just as efficient.


Round 3 - Honestly, I think I did well in this one, all things considered. But an hour after the interview, I realized that there was a really simple and clever solution to the problem that I missed.


Round 4 - I was soooo close to arriving at the optimal solution. I ended up needing a hint and then I implemented it towards the end. If I have any regrets about this one, I wish I just white boarded through an example, then the solution would've been obvious.


Round 5 - This was actually the behavioral (Googlyness) round, but my interviewer thought it was a coding round. I pointed this out and told them I was happy to do another coding round if needed, but then they realized they made a mistake and asked me a bunch of behavioral questions. Definitely the easiest round, even for someone with mediocre social skills like me lol.


Overall it was a good experience for me. I got another chance to interview with Google 1.5 years later and I feel like I crushed those interviews, and ultimately got an offer.

---

If you're preparing for interviews you might find my coding interview roadmap helpful - neetcode.io

There's also a bunch of courses with interactive lessons to teach you all the tricks i've learned!

#dsa #leetcode #codinginterview

I’ve solved a lot of Leetcode problems (about 665, and I made YT videos explaining 472 of them)

I was unemployed and had way too much free time. But as I solved problems I learned there are different levels of preparation.

🥉 𝗟𝗲𝘃𝗲𝗹 𝟭 (𝟱𝟬 𝗽𝗿𝗼𝗯𝗹𝗲𝗺𝘀)
✓ I was able to solve the easiest problems like reversing a string or inverting a binary tree, but I hadn’t even touched a graph problem.
✓ I realized that you do NOT need to know all the advanced DSA concepts before getting started. Knowledge of arrays, linked lists, hashmaps, and trees should be sufficient.


🥈 𝗟𝗲𝘃𝗲𝗹 𝟮 (𝟭𝟬𝟬 𝗣𝗿𝗼𝗯𝗹𝗲𝗺𝘀)
✓ I learned most of the high-value algorithms like DFS, BFS, sliding window, backtracking, etc.
✓ I switched to using Python because it’s just a lot easier for coding interview problems.
✓ I still wasn’t able to solve unseen problems. Later on I realized this is because most LC problems are sequential. The order you solve them in DOES matter.
✓ This is why I created a roadmap to help people solve problems in the best order possible - 🚀 https://lnkd.in/dccBt8MH


🥇 𝗟𝗲𝘃𝗲𝗹 𝟯 (𝟮𝟱𝟬 𝗣𝗿𝗼𝗯𝗹𝗲𝗺𝘀)
✓ I was able to go from 100 → 250 problems solved MUCH quicker than I was able to go from 0 → 100 for this reason:
✓ I stopped spending more than 30 minutes on a problem if I couldn’t get close to a solution. It’s just NOT a good use of your time to waste hours banging your head against a wall for a single problem.
✓ I was able to write algorithms like DFS, BFS and Binary search without giving them much thought - the same way most programmers can write for-loops without thinking much.


💎 𝗟𝗲𝘃𝗲𝗹 𝟰 (𝟲𝟬𝟬+ 𝗽𝗿𝗼𝗯𝗹𝗲𝗺𝘀)
✓ I can solve most medium questions in 10 - 15 minutes.
✓ I realized no matter how many problems you solve, you still won’t feel 100% prepared for interviews.
✓ Preparation is more of a distribution. You may have a 50% chance of passing an interview, or maybe an 80% chance. If you’re a perfectionist, I’m sure you want a 100% chance, but that’s just not realistic.
✓ There are diminishing returns on solving more and more problems.
✓ Quality > Quantity. It's better to have a deep understanding of a smaller set of problems, rather than a shallow understanding of many problems.


---


I’ve curated a famous set of 150 LC problems in this roadmap if you’re looking for a structured study plan: 🚀 https://lnkd.in/dccBt8MH

Imagine that you’re Notion. You just used SQLite to improve browser cache performance by 20%.

But you have a new problem.

Some users are experiencing worse performance. Why?

Well Notion was caching data locally. And believe it or not, for some really old devices it was actually faster for them to make a request to Notion’s backend, than it was to read from local disk.

This was only affecting a small number of users, so what would you do?

The solution is simple. Just have the browser “race” the two async requests. The application will use whichever request returns first and ignore the other.

---

If these concepts are new to you, you may be interested in my System Design for Beginners course: neetcode.io

Here comes the NeetCode system design arc. 🔥🧯

Follow for more!

I was solving a leetcode problem yesterday. It was a dynamic programming problem, and I was explaining the bottom-up solution.

I didn't even notice this until someone in the comments pointed it out, but it was identical to Pascal's Triangle.

Just a reminder to me how beautiful nature is.

I always say I love the work that I do. But do you really think I wake up everyday and say “oh boy, i can't wait to solve some leetcode problems today“.

Passion isn't something that you're born with, it's something that you cultivate.

Taking moments like these to just appreciate life is one of the things that keeps me motivated.

---

The problem was Target Sum btw - https://lnkd.in/gDXaMms3

I got a chance to talk with Matt Ranney who's an engineering leader at DoorDash.

We talked about:

- Microservices are actually just tech debt
- Developer productivity
- If microservices 'worth it' for DoorDash
- Programmers can be overly dogmatic
- Programming advice & more

I'm honestly surprised with how much it's resonating with my audience, given that it was a very deep & technical conversation. I thought some of you may also find it interesting.

Big thank you to Matt for taking the time to chat with me, I found it extremely insightful.

And sorry Matt for over-saturating your face in the thumbnail, that's just how YouTube works these days 😅

---

Here's the link: https://lnkd.in/gMUMYywX