YOE: 2+\nCurrent Position: CV focused MLE\n\nI don\'t see a lot of MLE interview experience in LC, so thought maybe I can give back (bounded by NDA still). At FB, MLE and SWE interviews have lots of overlap except there is **one extra ML design** round for MLE position.\n\nInitial contact by recruiter through LinkedIn. Did phone round in January, scheduled onsite in February.\n\n## Phone\n\nQ1: medium array problem\nQ2: medium tree problem. With follow-ups\n\nBoth questions can be found in Top 100. Solved both just in time, not enough time for question at the end.\nAfter this round, I focused on improving talking while coding.\n\n## Onsite\n\n### Coding 1\n\nQ1: medium 2D array/matrix problem, don\'t think its on LC, but similar variant can be found in LC for sure.\nQ2: medium tree problem. Can be found in Top 50/100.\n\nSolved both but struggled a little for Q1 because it was the first interview and question. Interviewer helped a little bit at Q1 to calm my nerves. Left about 5 mins for some questions.\nSelf eval: Lean hire/hire\n\n### Coding 2\n\nQ1: medium math problem.\nQ2: easy tree problem.\n\nBoth found in Top 50. Done both before so were able to solve both optimally. Also left about 10 mins for some questions, and ended interview a little early.\nSelf eval: Hire/Strong hire\n\n### SWE System Design\n\nFocus topic is quite common. Can be found in Gr\\*\\*\\*king system, YouTube etc.  \nI followed approach outline in gr\\*\\*\\*king. \nWe deep-dived into some components of interest. The lack of actual experience in system design probably showed here, as I struggled a little the deeper we go. Nonetheless, the entire interview felt like a conversation or brainstrom session with a co-worker.\nSelf eval: Lean hire\n\n### ML Design\n\nI was actually quite nervous about this round because I could not find a lot of resources for this and I didn\'t think I am experienced enough to wing this round. \n\nI was asked to design something similar to a rec/ranking system. The approach I took was broadly similarly to SWE system, but the discussion went for breath instead of depth. From data to model to experimentation to deployment and debugging. I think this was my weakest round. The interviewer did not seem interested or impressed throughout the interview. I knew this would be my downfall. \n\n**Moral of the story**: prep for rec/ranking ML system. CV problems are very different from rec/ranking. \nSelf eval: no hire\n\n\n\n### Behaviorial\n\nPretty straight forward. I preped this in a way similar to Amazon\'s LPs.\nSelf eval: hire/lean hire\n\n\n## Tips and Remarks\n\nThere was some chatter about Meta lowering hiring bar, but I personally do not think that\'s true. However, I do believe with the **right preparation**, experience and a bit of luck, Meta is one of the easiest to crack. May try again in the future for specialist role that is more CV focused.\n\nFor coding, prep for talking through code, and focus on FB tagged questions.\nFor system design, prep with gr\\*\\*\\*king, watch lots of YouTube (InfoQ, Gaurev, etc), read lots of blogs. Probably the most fun to prep for as it also directly translates to your work.\nFor ML design, prep for rec/ranking system for generalists. This was definitely not my cup of tea, and it showed during the interview.\nFor behaviorial, prep with stories related to their core values in STAR format.\n\nLC stats: ~400: (E) 25% / (M) 65% / (H) 15%

YOE: 2+\nCurrent Position: CV focused MLE\n\nI don\'t see a lot of MLE interview experience in LC, so thought maybe I can give back (bounded by NDA still). At F