Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
Inside Cursor
5 Plays11 days ago
Episode 2
A self-taught, "non-technical" guy ended up at one of the most important AI companies in the world. What does he see about the tool, about the work, about the people that rest of us are missing?
The Root Cause - A podcast by RubixKube
Learn More
Recommended
Transcript
Introduction to Mohit and Cursor
00:00:01
Priyank Upadhyay
hey everyone, welcome to the Root Cause. We have another guest today. And our guest works at Cursor, the AI code editor that has taken um developer community by storm and is one of the most talked pieces of software out there in the industry.
00:00:25
Priyank Upadhyay
Mohit runs user operations there and he joined just a few months back which means he sits with users every day and sees what people are actually building with Cursor.
00:00:36
Priyank Upadhyay
um And one very interesting fact about Mohit is he doesn't have a computer science degree. He never has been formally trained as an engineer.
00:00:47
Priyank Upadhyay
He is self-taught and he is inside one of the most important AI companies in the world right now. So welcome, Mohit. Welcome to the root cause. How are you feeling?
00:01:01
Mohit
Feeling great. Thank you so much for having me. Super exciting to be here and like talking about all things AI and Cursor. As my connection, like I'm part of like a around like pretty good, capable people like at Cursor. like I'm part of the user operations team. I am personally a technical support engineer.
00:01:20
Mohit
And as you rightly mentioned, it's like I don't have a computer science degree background, but yet I'm an engineer now, right, like on paper. So I think it's quite interesting. And it's just like, it's been super...
00:01:33
Mohit
inspiring and magical for me that someone like me also can make myself sort of technical without having those sort of backgrounds and it's incredibly uh interesting that companies are also open to these sort of like candidates and like open to like hire you on the basis of like qualification not just qualifications but like just merit and whatever you have uh learned on the in your career journey right like so for me that has been like super interesting but once again thanks for having me super excited for the chat you're going to have
00:02:02
Priyank Upadhyay
Awesome. Mohit, one thing.
Mohit's Self-Taught Tech Journey
00:02:05
Priyank Upadhyay
It's a very obvious question what my audience would be wondering about. You don't have a computer science degree and not formally trained.
00:02:14
Priyank Upadhyay
So how did it happen? Can you talk about your journey?
00:02:19
Mohit
Yeah, for sure. ah So I think, like again, a lot of things start early. right like So for me, I think back in when I was in ninth or 10th grade itself, I had this whole phase of, like oh man, this I got my first laptop and like I started playing Road Rush like any child should do.
00:02:34
Mohit
And then eventually it started going towards directions of, like okay, this is interesting. I can make a hello world HTML, which was taught to me in computer science class. right Then it goes to a point where my Android phone stopped working.
00:02:46
Mohit
And I was like, okay, i don't know. I took it to the service center. They had no answers, right? And that was the time when like stack overflow and all of those things were like up there. I was like, you know what? There's no way I'm buying another 6,000 rupee phone. There's no way my parents are gonna let me. Let me see if I can do anything.
00:03:01
Mohit
And then I connected my phone to the laptop. And then, you know, like all those, like at that time, routing was a thing. Like all those, like there were weird, sketchy sites where you can like kind of jailbreak your phone and like do very questionable things to them. right like And then that kind of got me into that. I was like, damn, I repaired my phone. And that was just by like going to some sites and storing some very weird SDKs. And it worked out. right like For me, I think that was the first time. And then somehow in the middle, I was just like, OK, maybe computer science now. I'm still very interested in this, but commerce seems like the way to go. And I was also naturally like into finance at that time.
00:03:37
Mohit
So felt like that was the right call. However, I think like the technical part of like those things, like the debugging and like just trying to find the... How much can you expand this technical limitations right of your device and kind of got me interested. And I've always... like Even when I was doing all these commerce things, I have like probably i have seven accounts on freecodecamp.org, which was like my go-to site to like start learning how to code back in the time. So I was trying to learn how to code, but it just was never either accessible enough for me, not easy enough for me. I used to get demotivated because I don't have the qualification, right?
00:04:14
Mohit
then then so But then something super magical happened last year where I was working with this company at the time called
AI Empowerment for Non-Experts
00:04:21
Mohit
Chomp. And then we were going to like ship some marketing campaigns and the engineering, of course, we had our bandwidth limits and they were focusing much more on the app and everything.
00:04:31
Mohit
There was like a need of like sort of a custom built dashboard, but for marketing and growth reasons. Right. And I was like, okay, I've been hearing a lot about these AI tools on Twitter and everywhere. Maybe I can try it.
00:04:42
Mohit
And then I started trying and oh boy, it was a surprise, right? Like I was like, it took me a couple of days. It took me like a couple of like good dollars of burning just to like get the basics right.
00:04:53
Mohit
But eventually cut to a week down, we were able to ship out like a entire microsite which had proper API connected to our back ends proper like, and basically any of our 5,000 users could go in there, check their standing in the leaderboard. So it was like a truly pivotal or like I would say magical experience for me that I have always imagined that I was limited by my let's say qualification or like just lack of knowledge to be able to ship something which is like you know sort of like to say production grade.
00:05:21
Mohit
So I think that was kind of a turning point for me that okay I think we are at a future where at least some important work can also be deliver delivered by using this so so so to say wipe coding ai tools right.
00:05:33
Mohit
And this was earlier last year. So since then I had like, of course, a natural inclination towards like, damn, let me see how much more can this go. And of course, as the model kept getting better, as the products and like all these tools got better for me, I started feeling like there there are very few limitations now, right? Like the probably like limitation is just imagination and my credit card limits, right? Like depending on how the tokens are going to charge me. So then, yeah, I spent like a good six months in just generally understanding ai Like I did try to understand the theory, but at some point I realized like, let me try to get the use cases and like, you know, deployments, right? Rather than like the theory itself. So probably if you still, even today ask me like, okay, what exactly happens when I put a prompt in, let's say GPT and what happens at the backend? I might not be able to answer that very well, but at the same time, okay, do I know how to prompt better? What kind of gets better responses and able to like,
00:06:28
Mohit
You know, help me with the work? The answer is probably yes, right? So that was the later half of last year where I was like, okay, experimenting with all this. And then I think somewhere in January, I was like, okay, I have got this AI fever. I really just need to, you know kind of be part of something which is at the forefront of something so revolutionary.
00:06:47
Mohit
And I used to use cursor a lot back in the time also.
Joining Cursor: From User to Team Member
00:06:50
Mohit
i In the starting, I found it like very daunting because, again, it was in a very IDE sort of a framework. It was like a lot of files, lot of context overload for me at that time.
00:07:02
Mohit
But then slowly, slowly, and as the app itself has progressed so much, it made me feel like this tool is super powerful. right like like Having more control can be incredibly powerful for me. and that's what inspired me that okay i love this company let me see if i can find my way in right so in january what i started doing was like trying to assess like what my skills are what my knowledge is what my experience is versus what could cursor potentially need and what do i like doing right like at that venn diagram of like five different priorities i found in the center like maybe something between the users and engineering sounds like good to me uh where i get to be part of the technical team but also It's not like on me to deploy some code.
00:07:43
Mohit
So I think I started spending a lot of time on Cursor Forum, which by the way, if people like here are listening, Cursor Forum is an incredible tool for people to find support. A lot of communities stays there and like kind of shares their workflows and help each other out. So I started spending a bit of time there and I was like, okay.
00:08:01
Mohit
this is what I need. Like I started going through a lot of these posts, made like a nice little deck sort of a thing out of whatever I think could be automated, could be better, how we can moderate this better.
00:08:13
Mohit
And then I just shot my shot with like a cold email to the CEO. And then i think eventually one thing led to another and they liked the whole oh analysis which I had done. Later I realized like most of it was already done in the company by the way, right? Like it was me from an outsider point of view, which who did not have enough context.
00:08:33
Mohit
But the whole point I think was that, okay, I was giving them something and they could see like this person is probably passionate about this. work itself.
00:08:44
Mohit
And yeah, that's how Cursor happened. They hired me earlier in February and it's been amazing ever since. It's been almost three months, but it feels like at Cursor we keep saying like it's been years, Cursor years. So three months have been like incredible, got to
Evolving Tech Industry Values
00:09:00
Mohit
learn so much. We have already pushed out so many new things. the Interface itself has completely changed from when I joined Cursor to now, which says so much about like the rate of shipping. right like When I joined, there was like a one single sort of like an ID. Now we have like ah another interface, like Agent Window, so and then Automations, and then Cloud Agents. So it's incredible. It's super.
00:09:21
Mohit
Yeah, so that's like my short, long answer of like how I reached here, where am I coming from, et cetera.
00:09:21
Priyank Upadhyay
It is.
00:09:27
Priyank Upadhyay
It is inspiring, Mohith, actually. A lot of folks wonder, if I do not have technical skills, will I be part of a technical setting in a company?
00:09:40
Priyank Upadhyay
I think things are changing drastically, and it opens up a different kind of vertical where your skills are more valuable than your degrees. Right?
00:09:55
Mohit
Absolutely.
00:09:55
Priyank Upadhyay
Yeah.
00:09:55
Mohit
i I mean, I think, to be very honest, I do want to believe in the fact that this was always the case. But I think with the, you know, like Twitter and all these incredible resources, we have to learn more and like how to find gateways to like people's inbox.
00:10:12
Mohit
I think it has become a bit more easier. Again, I would like to believe that this was always the case that degrees did not matter. But yeah, for sure, like, It has changed, right? Like people are not emphasizing enough on degrees, of course, in like some very specific domains and like there are niche circles where the degree is definitely helpful and still useful.
00:10:32
Mohit
But at the same time, I do feel like there is real opportunity here for at least as an individual, like even if you are not thinking from company point of view, as an individual, you should remove that limitation you have on yourself that, okay, I did BA economics in college. Can I like still not join some of these best technical companies in the world? I think that limit removing that limitation is definitely a step one, right? like Because we have access to all the best AI called coding models. We have access to just general LLMs who are able to teach us things. like
00:11:03
Mohit
You can ask it like how to build a rocket or something and be like, okay, now explain like I'm 15, then 10, then 5, right? Until you really get to a point of like, okay, I think I understand the basics now.
00:11:14
Mohit
So I think for... Anyone to have a limited mindset of my degree does not allow me to apply for this job is sort of like a bad thing. I really think like you should do a honest check with yourself that okay, it does not mean to say that if I see a CTO position at like the best company I apply there, right? But at the same time, if i see something which is like okay can i really learn it can i really excel in this job and if the answer comes to be like a yes then start training towards it then you do like the whole how do i benefit to them and show them my sort of like value before even joining them right like i think for me that was a forum analysis approach for different different companies it could be different right like i could imagine a world where it's like let's say i'm a marketing guy
00:12:00
Mohit
I would just go to your Twitter, i would check out your last 50 posts, I would be like, okay, how can I use like this hicks Hicks field and like the best AI video creation tools and give you like five media generator, media assets and say that, hey, I did this without you paying me, imagine what can I do if you pay me, right? Like for me, I think that proof of work sort of part of the application has become super important.
00:12:25
Mohit
ah if you don't have the necessary background or degree so to say. yeah
00:12:32
Priyank Upadhyay
Makes sense. Makes sense. Yeah.
Mohit's Proactive Workday with AI
00:12:34
Priyank Upadhyay
What does a typical Tuesday look like to you? more
00:12:40
Mohit
oh Yeah, I think it's interesting. A lot of my work always has to do around with like user queries. Like I'm part of the forum team. I'm also part of like the email support system we have ongoing and sometimes occasionally social media support. Right.
00:12:55
Mohit
So I think for us, a lot of it is like, OK, is there any, for example, like when I imagine daily day, it's like, OK, again to plug in cursor automations and cloud agents here, I have an automation running which goes through our company Slack channel, which goes through like a couple of other dashboards and just gives me like a daily digest, which is like the first thing I see every morning that, okay, has there anything been like significant since I went to bed last night, right? Like that did we have any sort of like a release? Did we have some sort of a policy change? Did we have some sort of like a,
00:13:30
Mohit
could be just error rates in particular part of the system, whatever, right? Like, so that just preps me for the day. And then I go about my day by like looking at all these different customer portals we have, like the email support and like forum and just like trying to unblock the user as much as possible, right?
00:13:45
Mohit
A big part of being a technical support engineer and cursor, I used to think that would be, okay, I replied to this customer and it ends there, right? Like, or I tried to like file a linear bug and it ends there. But I think it's been incredibly important powerful being in this position because we are not limited at this point by just like reply to the consumer and reach out to the engineering know if you want and if you have the idea skills you are free to like ship whatever workflows you are free to ship whatever automations you think will make the job better there is no unnecessary
00:14:18
Mohit
uh approval matrix of sorts right a lot of like culture is like okay why don't you ship this and we'll see how it goes as long as we are very sure that this is not impacting our customers right at least for internal purposes we are very like uh supportive of that sort of a culture so for me like what usually can involve like i'd get done with that part of my job which is to deal with the customers and then it's also like okay am i identifying any sort of general patterns today which i did not find the last week or like yesterday right And then can I do something to help this make better? For example, to if I'm seeing an increased sort of ah error rate from like, you know, okay, these particular users from ah this particular network configuration are not able to use cursor as well.
00:15:04
Mohit
So, okay, what can I do here? Of course, one idea is go engineer and be like, hey, can we fix this? But again, I think that is... like just step one right like of what you can do as at cursor or even if any other company who uses like one of these ai tools or even cursor with like what you can do is like what i would do if i see this now like i'll ask cursor that okay why is this happening do you see a common pattern across these multiple user reports and then cursor like how about you ask them to like run our network diagnostics right like just to get that fixed piece fixed
00:15:35
Mohit
And then I'll do that. And then maybe the consumer will reply, no, it all seems fine. So now we have ruled that out. It's probably not like this has happened. And then at one point we identified like a particular antivirus software was just doing like a minor change to something we were doing on the IDE setting.
00:15:53
Mohit
And that was like blocking a couple of users. Right. And we did not need engineering help to sort of identify that. Right. Like it just reduces their bandwidth at the same time, ah same time making us feel so powerful that it's just such a It used to be conventionally such a big blocker that every single query has to go through our engineering eyes and like then get reviews and then probably a hotfix will get pushed or something.
00:16:14
Mohit
But now we are at a world where it's like, okay, if you are confident about what you have identified, just start communicating it out to the users. Let's unblock them as fast as possible. right So I think for me, ah yeah the three different parts are like user queries, workflows, and then just debugging on like identified patterns.
00:16:34
Priyank Upadhyay
Makes sense. This has, a the shift has been amazing actually with the use of agentic tooling and the overall explosion of AI that has happened past two years have drastically changed a lot.
00:16:51
Priyank Upadhyay
um ah Roles are being, you know, Question now, whether or not some roles are actually relevant anymore.
AI's Impact on Job Roles and Opportunities
00:17:00
Priyank Upadhyay
do Are you also seeing that shift ah in terms of who will be responsible to do what or is it all getting mingled now?
00:17:10
Mohit
Oh, well, I think, first of all, I think it's important to acknowledge that definitely AI has been shifting the industry, right? Like to be on the either side that AI has completely transformed and we don't need employees anymore. And to the side where, nah, we still need everyone, right? Like, I think I am somewhere in the middle where it's like, here different yeah, definitely AI has impacted like how we are seeing jobs now or how hiring has been taking place.
00:17:33
Mohit
And I think it's incredibly like, like it would be just very, ignorant to say that AI has changed nothing. right For example, like copywriting, like one of those things, right? Where maybe a company used to need 50 of them for their different push notifications. And that's probably not done by 50 people anymore. And it's by five people, right?
00:17:52
Mohit
Probably even like my personal role, which is like a support role. A lot of companies probably have very intelligent chatbots, very intelligent systems in place. And then that reduces like you needing, like, let's say one thousand, thousand support people to like, let's say 200 people. Right.
00:18:06
Mohit
So one, like, yes, there's definitely a change in how the hiring or like just the, uh quantity of hiring which has been taking place right but at the same time i do feel like there's been a lot like it's sort of like i do believe in the javance paradox in the sense that yeah when we increase some capabilities there's also an increased output and which also means increased consumption which also means like increased need of human labor in that sense so i would like to be cautiously optimistic and be believe a believer of that uh but i think like
00:18:40
Mohit
Yeah, to answer that, I would say hiring has changed, but we still do need a lot of humans in the loop. There's a lot of judgment. There's a recent conversation about the whole taste. So I think that matters. I have made a council of AI agents myself to somehow steal man a certain idea.
00:18:59
Mohit
still man a certain idea But like at the same time, at the end of just sometimes the day, sometimes answer is like, h okay, AI told me to do this. like Seven LLM models fought against each other to give me this idea.
00:19:10
Mohit
But I still don't think that's the best idea. right I would do like a different thing. Maybe it's inspired from what LLM suggested, but it will be a different thing. So I do feel like there's always...
00:19:18
Priyank Upadhyay
right.
00:19:19
Mohit
like some part of human judgment to be played in things like this because if you ask AI it would be like yeah why don't you make your site into like 200 languages in the world so that every SEO is like optimized and like oh everything is like you know but at the same time is that the best decision business decision for you to make we don't know right like so I think that's how I see it like Yeah, like the hiring itself also definitely has changed in the sense that you are expected to do more.
00:19:48
Mohit
But I think that's a good thing, right? Like it does not like if I have the tools, if I have the skills, if I have the knowledge and I'm not limited by it, i don't think it's unfair for the employer also to be expecting a bit more, right? Like you can't be held like you can't expect to be not held accountable to where the world is going right like that would be like you bring a typewriter to a company who's asking you to get macbooks right like because again you believe in typewriters sure but like then it's better off if you find a typewriter kind of a job right like but at the same time you just gotta stick with it and run with it alongside yeah
00:20:25
Priyank Upadhyay
Makes sense. Do you think there is a limit where
AI in Decision-Making: Efficiency vs. Judgment
00:20:30
Priyank Upadhyay
ai falters? I'll tell you one instance that personally happens with me is I used to write you know three pages when I had to write an email.
00:20:41
Priyank Upadhyay
First a rough draft, second a refinement of that and then a final draft. Then I used to send it. now I am getting so lazy that I you know use any kind of AI tool to help me generate the emails.
00:20:57
Priyank Upadhyay
Do you think it's getting to a point where we are using ai for thinking itself? Which is a like the most appropriate human trait that we should be taking forward.
00:21:11
Mohit
Again, I think like we bowed down to the algorithm algorithmic overlords long back to even discuss this now. right like If we are thinking that it took LLM for us to stop thinking, I think like we kind of like started decaying in that thought process long back even before the LLMs came in because just like i can like all these social media which kind of influences how we think and all of that. right like So firstly, like I think we have already been influenced to like I don't know, not think so much from first principles, not think your original thoughts. You think it's original, but you probably got an idea from somewhere else, right? like So I think the programmatic like evolution of that thought process started like a bit before that. But yeah, for sure, like I do think that
00:21:56
Mohit
we have like for me also like personally like first thing i do as soon as i receive an email or anything complex is like okay this is what i think but let me also like go through an llm and like kind of see what it says about this right and then i take a judgment call of like okay which is better and i would be lying if i say that somewhat sometimes the i just copied and like what the llm gave me i put it in the right but at the same time it's like there were enough and more examples of it being the other way around too, right? Like where I saw the LNM response and I was like, nah, this does not sound good, right? Like there's no reason I have to give them a 40, like 40 line answer when I can just tell them like a simple answer, right? Like, so I think the whole taste judgment still creeps in, but to answer your like point shortly, I do think it's gonna make us...
00:22:47
Mohit
lazy but also efficient right like so it's like one of those weird trade-offs that okay is it's making me lazy but also it's making me so efficient right like so i think at at the end it just comes down to you as a person like kind of being my mindful about like okay how much of it am i making it do and am i still like am i being a kind of just a blind like Am I its employer or the employer? right Like the LLMs. So I think that kind of a thing. like Am I just listening to what it asks me to do and I just do it? Or am I training it to like be my assistant that, hey, like make this better. No, you did not get this right. So I think that sort of a mental shift is definitely important.
00:23:25
Priyank Upadhyay
Awesome.