Learning Library

← Back to Library

Hiring Manager's AI Resume Rules

Key Points

  • Do not let a large language model write your entire résumé, because its default “house style” will make you sound generic and blend in with other applicants.
  • Avoid using an LLM to answer interview or application questions, as the responses tend to be vague, word‑y, and fail to showcase the clear, incisive thinking recruiters look for.
  • Refrain from using an LLM to “tune” or keyword‑pack your résumé for each job description, since this strips away your personal voice and often inserts keywords in the wrong places, making the document look mechanical.
  • Only consider AI tools if you’re skilled at crafting system prompts that fundamentally change the model’s style to match your unique tone; otherwise, the AI’s inherent style will dominate and hurt your chances.

Full Transcript

# Hiring Manager's AI Resume Rules **Source:** [https://www.youtube.com/watch?v=iP8p7YjGG9k](https://www.youtube.com/watch?v=iP8p7YjGG9k) **Duration:** 00:14:36 ## Summary - Do not let a large language model write your entire résumé, because its default “house style” will make you sound generic and blend in with other applicants. - Avoid using an LLM to answer interview or application questions, as the responses tend to be vague, word‑y, and fail to showcase the clear, incisive thinking recruiters look for. - Refrain from using an LLM to “tune” or keyword‑pack your résumé for each job description, since this strips away your personal voice and often inserts keywords in the wrong places, making the document look mechanical. - Only consider AI tools if you’re skilled at crafting system prompts that fundamentally change the model’s style to match your unique tone; otherwise, the AI’s inherent style will dominate and hurt your chances. ## Sections - [00:00:00](https://www.youtube.com/watch?v=iP8p7YjGG9k&t=0s) **AI Resume Writing: What to Avoid** - A hiring manager explains why applicants should not let large language models fully craft their resumes, warning that the AI's generic “house style” can erase individuality and make candidates sound indistinguishable. ## Full Transcript
0:02we need to talk about resumés and AI 0:05because there are a lot of people who 0:07are giving really conflicting 0:08perspective and I'm a hiring manager 0:11I've been hiring in this cycle and I 0:13have fairly strong opinions about where 0:15AI can effectively be used by applicants 0:19and where it should not be used so I 0:20want to break that down into two lists 0:22and I'm going to go into a fair bit of 0:23granular detail that's what a nice long 0:26form video like YouTube is about so 0:29first a good list of things that you 0:31should not be using a large language 0:34model for during your job application 0:36process are you ready number one do not 0:40use a large language model to write your 0:43whole resume for you don't do it the 0:46reason why is that large language models 0:49have a house style unless you are 0:53reasonably a Adept an expert at 0:57prompting large language models with a 0:59system prompt that substantially changes 1:01their house style so that it sounds very 1:04very very much like you and it's 1:06distinctive and it's individual then 1:08don't do it and if all of that didn't 1:10make sense to you then you're not using 1:12a system prompt that is going to help 1:14you and you should definitely not be 1:16doing it fundamentally unless you have a 1:19fair bit of work put in an llm is going 1:22to sound like an llm Claude has a house 1:26style Gemini has a house style 1:30open Ai and chat GPT have a house 1:33style do not let that house style into 1:36your resume because so many other people 1:38are doing so you end up sounding like 1:41all the other applicants is that what 1:43you want is that what you're looking to 1:45do is to not stand out from the crowd 1:48because that is typically what happens 1:51when you let an llm write your resume 1:54all right that's number one number two 1:57if there are questions from the company 2:00to you during the applicant process 2:03don't use an llm to answer them same 2:07reason if you use an llm you have a 2:09house style and for for a second reason 2:12those questions are typically designed 2:14to measure incisive and careful thinking 2:16and next token predictors which is what 2:18a large language model is are not going 2:21to give you the clarity of thought that 2:23you need to stand out it's going to feel 2:25like word salad maybe not inaccurate 2:29word salad but word salad 2:31anyway and if it feels even a little bit 2:34fudgy or a little bit flabby or a little 2:37bit sloppy then the person reading the 2:40response is going to think that's the 2:41way you communicate that you're not 2:43incisive that you're not concise that 2:45you're not clear and you don't want that 2:47again it's not helping you stand 2:49out okay third 2:52thing do not use an 2:56llm to tune your resume 3:00I know it's really tempting when you 3:02have a lot of applications to do to just 3:04say to the llm please tune the resume to 3:07match this job description but the 3:09problem is you lose all of your own 3:11style in that process and the house 3:13style comes back and the llm isn't 3:17necessarily inserting the keywords that 3:19you want it to insert on the correct 3:21bullets and if you are reading several 3:24hundred resumés you start to get an ear 3:28for when 3:30people are using keyword packing and 3:33llms tend to 3:35be not super subtle about hiding the 3:39keyword packing and that's actually one 3:40of the things you have to do to write a 3:42good resume yes you want to match the 3:43keywords but it needs to feel 3:46natural and in my experience an llm 3:49written resume has trouble adjusting and 3:53tuning and bringing those keywords in 3:55without making it reasonably obvious 3:57what's going on and it does stand out if 3:59you've read a bunch of 4:01resumés all right number 4:04four don't use an llm for the Final 4:09Phase of preparation for interviews and 4:11I'll come back to this theme but at the 4:13end of the day the reason why interview 4:15preparation is so important is because 4:16your brain needs to actually speak 4:18something out to sound 4:20natural so don't use it for interview 4:24prep or you're going to sound like 4:26you're reading from a large language 4:28model output you're going to sound 4:29really robotic and yes I know there are 4:32some people who literally get the llm 4:35answers printed onto a screen while 4:38someone is interviewing them and it is 4:40obvious and people do know and it's a 4:42huge red flag because we need you to 4:44think creatively and think incisively 4:46and think for yourself and so if you are 4:49practicing with an lolam you are 4:51unlikely at the final stage of practice 4:54to be able to articulate your answer in 4:59your style using your words and your 5:01voice in a way that feels natural so 5:03that Final Phase of interview practice 5:05has to be with another human has to be 5:08with a mirror if you don't have another 5:09human to practice 5:11with okay and then the last one that I 5:14would say that you don't do is do not 5:17use a large language model that has 5:19search capabilities such as chat GPT or 5:22even 5:24perplexity for company research that is 5:27high fidelity so if you're doing 5:29interviews and you get to like the 5:31second round or Beyond you need to have 5:33a really good idea of what the company 5:34does so that they can imagine working 5:36with you because you have so much 5:37corporate knowledge right 5:39already it is hard to get that level of 5:43uh tactile fine grained detail if you 5:47are using an llm as an interface and the 5:51reason for that is that they are 5:52designed to summarize and when they 5:54summarize they do lose detail and you 5:56need that detail to stand out from the 5:58crowd so don't use them for that really 6:01high fidelity research that you need to 6:03do to truly stand out in subsequent 6:05rounds of interviews okay that was a 6:07list of five different things you should 6:10not be using large language models for 6:12but there are things you should be using 6:15large language models for in the 6:16application process and that's the 6:18second list and that's what we're going 6:19to get to now so what you should use 6:22large language models for is one I think 6:25they're great tools for rfit analysis if 6:29you to think about does this role fit my 6:32skills does this role fit my job 6:34description is there a way I can 6:36quantify that or measure that yes there 6:37are pre-built tools like teal that do 6:39that you can also prompt and work with a 6:42large language model just in the chatbot 6:46experience and get a great result just 6:49by having them read the two documents a 6:51job description print out and then your 6:52resume and starting to have a 6:54conversation about the fit and the 6:56weaknesses and the gaps that's a super 6:57useful way to use a large language model 6:59to understand whether you fit a role 7:02another useful way to understand an llm 7:04this is number two get them to tell you 7:07what is distinct about that role that 7:11doesn't fit the other standard job 7:13descriptions again you're asking the llm 7:15to look through its enormous experience 7:17I guarantee you it has read more job 7:19descriptions in your job family than you 7:21have ask it to look at an individual JD 7:24and say what is distinct about this and 7:26then press it and make sure that it's 7:28really clear that's a great use for a 7:30large language model because that will 7:31help you to read and understand intent 7:34because the large language model can go 7:35beyond just saying this is a distinct 7:37sentence it can tell you this is a 7:39sentence and this is my supposition 7:41based on my extensive Corpus of text the 7:44whole internet of text on what the 7:47intent behind that bullet was and it may 7:49not be 100% right but it's right enough 7:52that it will give you a clue that you 7:53might not pick up on otherwise so I 7:55think it's a good use just to like 7:57understand what's distinct about resumés 7:59and help you analyze them rapidly or 8:02understand what's distinct about job 8:04descriptions because those are the ones 8:05you're looking 8:07at so number 8:09three it's really helpful to identify 8:12missing 8:13keywords and this is goes in the same 8:15category of like comparing r fit and so 8:17if if you're comparing r fit which is a 8:19different thing you have to understand 8:20at a high level if the if the rolles fit 8:22that's number one number two understand 8:24what's distinct about the role we just 8:26talked about that number three if you're 8:28going to apply do a second level of 8:30analysis between your resume and the job 8:32description and specifically look for 8:34missing keywords and this is different 8:36than keyword packing it's the first step 8:38it's the analysis step where it calls 8:40out the missing keywords llms do a great 8:42job at that it may be on you to work in 8:44the keywords but llms can help you pull 8:47them out and understand what you have to 8:49add in really 8:51fast okay number four llms are really 8:57good at helping you with blank page 8:58syndrome so you know how I said earlier 9:00in this uh in this YouTube that you 9:03should not use a large language model 9:05for writing a whole resume that's true 9:08you should also not use it for those 9:10questions I talked about that but if you 9:12need a draft it's a great drafter it 9:15gets you past staring at a blank page 9:17and that can accelerate you because 9:19there is nothing harder as a writer than 9:21just looking at the blank page and 9:22wondering where to start we just don't 9:25have that problem which is a 10,000 year 9:28old Problem by the way it's disappeared 9:30in 2024 we don't have blank page 9:32syndrome anymore or we don't have to 9:35anybody can get off of the blank page 9:37immediately and you should and you 9:38should use a large language model for 9:40that and then you should edit the heck 9:42out of 9:43it all right number five uh you should 9:48be looking at how you can break through 9:52fluency blocks in your overall 9:55application process with a large 9:56language model and I'm going to give you 9:58two examples oftentimes people don't 10:01understand networking and they and they 10:02talk about hey how do I Network how do I 10:04Network in my situation for my job role 10:07how do I get past the 10:09shyness large language models are not 10:11going to give you personalized Insight 10:14from someone who's done it before but if 10:17you're getting past the blank page 10:19syndrome for a skill like networking 10:21they are super helpful to give you a 10:23sense of what's in the box and 10:26especially if you prompt them and ask 10:28them to think about about it from a 10:30strategic perspective and a 10:31non-transactional perspective which is 10:33how you should be networking they're 10:35going to give you really thoughtful 10:37initial advice like a first passet 10:39advice if you're looking to understand 10:41the skill in general it can be helpful 10:43and almost everyone has gaps in their 10:48skill set over the course of a job 10:50application process there are areas 10:51where you are strong maybe you're really 10:53strong at resumé editing in areas where 10:55you're weak maybe that's networking 10:57maybe that's looking at other job sites 10:59besides LinkedIn and so what you're weak 11:01at is actually getting the full top of 11:03funnel for job Discovery llms are that's 11:06another thing they're good at like you 11:08can talk to them and say I don't know 11:09where else to go for jobs besides Linkin 11:11can you help me find some that is a good 11:13use for something like 11:15perplexity basically you want to tune 11:18the llm to give you a sense of how you 11:21break through that skill Gap it's not 11:24going to give you a high fidelity 11:26excellent super professional final 11:29answer on these things but it will get 11:31you through that initial really um 11:36painful feeling skill learning phase 11:39faster and that's what you want it's 11:41accelerating you and that's what you 11:42should be looking 11:44for okay and I also know at the top of 11:47this YouTube video I talked about how 11:49you should not use large language models 11:51for the Final Phase of interview 11:52practice and that is true you should not 11:54be doing that but you should be using 11:58large language models if you were 11:59feeling stuck and need to practice your 12:02initial interview structure and so if 12:04you're doing the first half of interview 12:06prep where you're determining do I have 12:08the anecdotes do the anecdotes match the 12:10anticipated questions do the anecdotes 12:13have the right structure to them so that 12:15someone who is a typical interviewer can 12:18understand what I mean am I using the 12:19star technique am I using the parade 12:22technique am I opening myself up to 12:24followup questions that are risky llms 12:27are great as conversational tools to 12:29help you get through that initial phase 12:31of interview practice so absolutely use 12:34them for that first half and then when 12:35you actually need to feel natural and 12:36practice it verbally that's when they're 12:38not helpful anymore all right I'm going 12:42to give you one final tip I believe this 12:43is tip number seven for how to use large 12:46language models in application prep and 12:49sort of the whole job search process you 12:52know how I said earlier in the video 12:54that you should not be using a large 12:55language model such as perplexity or 12:57such as chat GPT with 13:00for researching at a high level of 13:02fidelity your company that you're 13:03targeting great don't do it but you can 13:07use a large language model to get a 13:10highlevel view of the industry and the 13:12competitors for the company because 13:14again that plays to the system's 13:16strengths it allows it to look across 13:18the whole industry I guarantee you it's 13:20read more articles than you have and to 13:22synthesize it up and if you're looking 13:24to synthesize and summarize a lot of 13:27detailed information that you found that 13:28maybe not 13:29publicly available on the internet maybe 13:31you maybe you've dug in the sec's 13:33website and you found 10K filings for 13:36these companies and they're recent so 13:37they're outside the training data set 13:39for the llm throw them in it's going to 13:42be able to summarize them very very 13:43rapidly and at a good enough degree of 13:45fidelity that you can understand what 13:46the competitor strategy is pretty 13:48quickly and that is going to give you a 13:50working knowledge of the industry and 13:53competitors that makes you stand out 13:54because typically you don't see someone 13:57dinged in Loop Fe back sessions for not 14:00having a high enough Fidelity view of 14:02competitors what they're dinged for in 14:04that area is not knowing competitors at 14:06all or not knowing the industry at all 14:08and large language models can get you 14:09past that so there you go that is a 14:12series of five ways not to not to use 14:16large language models and seven ways to 14:18use large language models as you're 14:20going through your application and your 14:21job search process AI is a powerful tool 14:24for helping you just need to use it in a 14:27way that supports your overall journey 14:29and your ability to stand out and that's 14:31what I wanted to call out here let me 14:33know what I missed in the comments