From Research to Researcher

Hi everyone. I’m very happy to be here to share some thoughts with you. The title of this talk is “From Research to Researcher,” and the three keywords are action, understanding, and mindset. From the title, you can probably tell that I mainly want to talk about two questions: how to do research, and how to become a researcher.

Getting Mentally Ready to Start Research

Let me start with the first question: how to do research. Before you begin, I think there are a few things you need to work through in terms of your mindset.

I’ve noticed that many students feel intimidated before they start doing research. You feel that research must be difficult. You doubt your own ability and think you’re not capable enough to do it well. Or you think your foundations aren’t solid: there are so many things you haven’t learned, so much knowledge you haven’t mastered, so how could you do research well? You might also be afraid to contact a professor and ask to join their group.

I’ve seen this kind of self-doubt, or this tendency to dismiss your own abilities, in a lot of students. But where do I think that fear actually comes from?

I think it comes, fundamentally, from a vague sense of what research is, or from not knowing what it involves. You haven’t done research before. You don’t know what the whole process looks like, so you don’t have a clear picture of it. People tend to fear the unknown, and that’s how this feeling of intimidation arises.

But what am I really trying to say? Research is always about learning as you go. We all start by getting our hands on something. You begin doing it, and then, as you encounter things and discover what knowledge you need, you go and learn it. You learn while doing, and do while learning.

So getting started really matters. Action is the first keyword in the title. Your actions matter. You have to take that first step.

Then you learn whatever knowledge you need. That’s very different from taking classes and preparing for exams. Studying for classes and exams is usually a bottom-up process, while research is a top-down process.

What do you do when preparing for an exam? You work from the bottom up. You have to learn all the material that might come up in the course before you take the exam, so that you might get a good score.

Research, though, starts with something you want to do, a goal you want to achieve. Once you begin, new branches open up as you encounter things. Then you learn the knowledge needed for that particular branch. You learn what you need when you need it. That’s one point I want to make: there’s always more foundational knowledge to learn, so you need to get past that fear.

The Full Process of Publishing a Paper

Next, let’s talk about how to do research: the full process of getting a paper published. I’ve laid out a fairly complete workflow here. First, you need a general direction. You choose an area you’re interested in, or your advisor gives you a direction.

Then you start surveying that area. That means reading a lot of papers in the field. In the third step, you settle on a topic, a specific research question. This step is very important, and I’ll come back to it later. A good question essentially determines half the value of your paper.

Once you’ve identified a research question, you can propose a fairly naive idea. It can be very simple, just a straightforward method. Put it out there and allow it to be imperfect. You don’t need to come up with a perfect idea right at the beginning. That’s not very realistic.

So you can start with a naive idea. Then you need to establish the smallest feedback loop you can: propose an idea, run experiments, and test it. That’s the next step, experimental validation. Try things and fail quickly, so you can find out whether your idea works.

Of course, if you’re very good, or very lucky, the first idea you propose might work. Then you can jump straight to step seven. If it doesn’t work, you move to step six and enter a loop.

Why doesn’t it work? You need to analyze the failure cases. This is one of the most important steps. How should I put it? Failure is actually the norm in research, but the knowledge you gain is often hidden in those failure cases.

The development of a whole field works the same way. As the field reaches its current stage, what difficulties has it run into? Why do those difficulties arise? People analyze them.

That’s how the field moves forward. I’m talking about the development of a field here, but the same is true for an individual paper.

For example, on my previous project, I worked with my advisor. Whenever I gave an update at our group meeting, I tended to focus on the results: the numbers and metrics from my experiments. When he saw that the numbers weren’t good, because the method wasn’t very effective or wasn’t really working, he’d always ask me why. What was causing the problem? And I often couldn’t answer, because I hadn’t done a deeper analysis.

One thing he kept saying has really stayed with me: “Look into more details.” He wanted me to examine more details and do a deeper analysis. My project was about defenses, and one of the key metrics was the attack success rate. If your defense is effective, the attack success rate should be low.

So, for that metric to look good, the attack success rate needs to be low. If it doesn’t look good, it means there are attack samples that we failed to defend against. We were doing defense research, and some attacks got through. But I stopped there. I presented the metric without taking the next step and asking: why didn’t the defense stop them?

That’s why you need to analyze those specific attack samples. What are they like? What features do they have? What characteristics do they share? How are they different from the samples you successfully defended against? Analyze them, produce more specific, detailed information, and look at it again. Then you can understand why your idea doesn’t work.

Then you can improve it. After making the improvements, you run the experiments again and repeat the process. In fact, about 80% of your time goes into this loop.

But the real highlight of the research process isn’t the moment when your idea finally works, even though that moment certainly feels great. What’s more important is the observations and insights you develop while analyzing those failure cases. That’s the most valuable knowledge you gain from doing research.

These are things only you know. At that point, nobody else in the world knows them. That’s real knowledge. And in major companies or top labs, a lot of the know-how, some of the most crucial knowledge, accumulates through the experiments of the engineers, researchers, and scientists doing the work on the ground.

Once your method works, you can start running experiments at scale, write the paper, and submit it. After submission, there may be a rebuttal, followed by revisions, and, if things go well, acceptance. That’s basically the full process.

Of course, there are also large research groups, or what you might call paper production lines. When you join, your advisor or the senior student guiding you may already have decided on a topic. They might even have given you the basic idea. So you just start coding, run the code, do the experiments, and keep going from there.

In some cases, then, you start directly at this stage. The earlier part is where you think for yourself and come up with ideas. In the later part, you just need to execute. That’s the first keyword in the title: execution, your ability to get things done.

Of course, this is just one standard workflow, for a very common type of paper. There are other kinds as well. You might discover a new phenomenon, or a counterintuitive result, and then need to do more empirical studies or theoretical derivations to establish that the phenomenon or result holds more generally. That calls for more analysis, whether theoretical or experimental. That’s a different kind of work.

But the kind I’ve laid out and used as an example is the most common. It’s probably what most of you will encounter.

Reading, Experiments, and Writing

This process involves three technical skills: reading, coding, and writing. Reading means going through a lot of papers when you’re surveying the field.

One thing worth mentioning is that AI can now help with all three of these skills. For reading and literature reviews, you can use these tools. Many tools, such as GPT and Gemini, have Deep Research features, so you can ask them to survey the literature for you. But you need to keep one thing in mind: the results aren’t complete, and they may miss some very important papers.

That’s why, when I first start a literature review, I do the searching myself. When I was working on my first project, just starting out in research, I read all the papers from the previous five years in that area that I could find on Google Scholar and that had appeared at top conferences.

After reading them, you feel that you understand the field reasonably well. This step matters because the point of surveying the literature is to develop a basic understanding of the field.

Then there’s reference management. There are plenty of tools for that now. Zotero is a common one, and there are others.

You also need to take notes when reading papers, and there are plenty of note-taking apps. Here’s one thing I’ve learned: you don’t need to optimize your tools to the absolute limit. Many apps have all kinds of plugins that can improve efficiency, but you don’t need to feel that you have to try every one. That can waste a lot of time.

Some of the really impressive researchers I’ve met keep things very simple. One professor, for example, puts all his notes and research material in Google Docs. What does that amount to? It’s like having a single Word document where you take notes, put everything down, and organize it. It’s a very simple setup.

But it’s also very efficient. If you’re used to it, it works. I just wanted to mention that.

There are so many papers now that you don’t necessarily need to read every one. You can divide your reading into two categories: skimming and close reading. For skimming, you can ask AI to read a paper for you. Afterwards, you have a general sense of what the paper is about, what it did, and what it contains. That’s enough for that purpose.

But what’s more important? You need to build up the number of papers you’ve read closely. There are many important papers that you simply have to read yourself. I’d suggest not using AI during close reading—although, of course, you can use it. What I mean is that you need to go through every sentence yourself. You shouldn’t have AI read it for you and then skip reading it yourself.

The second skill is coding. As I said, AI can help you write code, but the key is that you must be able to understand what it writes. You can’t just have AI generate something, run it, see that it runs successfully, and then trust it completely and start using it. There may be logical errors in there, which would make your experimental results unreliable, right? At the very least, you need to have control over the whole project.

For experimental data, one very important lesson is to record and organize it promptly. Once you start running large-scale experiments, the volume can be enormous. That’s especially true, for example, when you’re working on a paper for one of the four top security conferences.

Sometimes I’ve forgotten to organize my results, and a few days later I can no longer tell which experiment a set of data came from. So you really do need to record and organize things as you go.

When it comes to experimental design, there’s an underlying principle: all your experiments are there to demonstrate that your method really works. You’ve found that it works, and now you need to show readers and reviewers that it’s genuinely viable and effective.

You can run all kinds of experiments and extend them to different settings: test lots of models, compare against lots of baselines. All of that is there to demonstrate that your method is effective.

The third skill is writing. This also explains why I said you need to build up your close reading. Writing really draws on the reading you’ve done.

First, there’s the structure of the paper. You can look at good papers in the same field, see how they’re organized, and use them as a rough model for your own structure.

Then, at the level of wording and phrasing, the difference between a well-written paper and a poorly written one is very noticeable when you read them. You need to learn from the well-written ones.

In those papers, there often isn’t a single wasted sentence. First, you need to understand what each paragraph does. Then you need to understand that each sentence has a role within that paragraph, too. Think about those questions, learn from how the authors do it, and use that experience in your own writing.

Generally speaking, a good comment you can receive about your writing is “well written” or “well structured.” If a reviewer says that, it means the writing is in good shape. That basically covers what I wanted to say about how to do research.

Becoming a Researcher and Developing Research Taste

Now we move to the second part: how to become a researcher. First, what matters most for a researcher? Personally, I think it’s research taste.

Why does it matter? Think back to execution, which I mentioned earlier. To get a paper published, execution can actually be enough. You just have to do the work, keep at it, practice, and repeat the process many times. Then you can get a paper out.

But what does taste determine? Execution gets you a seat at the table, while taste determines whether you can stay at the table. As a researcher, you need to keep producing reasonably high-quality work.

That’s where your taste really matters. Let me first explain what I mean by taste. You can think of it as a sense of what makes good research, or a kind of academic aesthetic judgment.

What kind of ability is it? It’s the ability to choose what problems to work on. There are thousands upon thousands of possible research questions. Which ones do you think are worth studying? You choose those. It’s a matter of choosing, judging, and deciding: what do you think is valuable?

How Do We Define the Value of a Paper

Let me insert a few personal views here on how to define the value of a paper. Everything I’m saying in this part is my own opinion.

What do I think research is fundamentally about? It’s about communicating something you want to tell the community. A paper’s value lies in how much that information adds to the community’s understanding. How much does it bring that people haven’t seen before? That’s the new understanding it contributes.

For a researcher, I think the greatest thing is to discover new knowledge.

And then to turn that knowledge into common knowledge. Think about the courses you’re taking. The knowledge in your textbooks was new when it was first discovered, but now it’s all common knowledge.

The people who discovered the knowledge in our mathematics and physics textbooks have gone down in history. They’re great scientists, and I think that’s the greatest thing a researcher can do.

Let me explain this idea of adding to our understanding a little more concretely. Imagine the community’s response. The most impressive kind of work is when you solve a problem that has remained unsolved for decades. People react by saying, “Wow, this problem can actually be solved!”

For example, last year, the mathematician Hong Wang solved the three-dimensional Kakeya conjecture. When was that conjecture proposed? In 1917, more than a hundred years ago. A problem that had remained unsolved for over a century was finally solved. That’s tremendously striking: it takes the world by storm and makes you internationally known. And Hong Wang is now also a leading contender for a Fields Medal.

Of course, mathematics has more of these old problems. In computer science, we might be talking about problems from the past few decades. If you can solve a problem that nobody has solved for decades, then even if your writing is terrible, that’s okay.

It doesn’t stop your paper from having a huge impact. That comes from the problem you’ve chosen to study. The problem itself is very important.

The second kind is when you propose a new paradigm and it works quite well. People think, “Oh, so you can do it this way!” That’s work that gives people new ideas. Proposing a new paradigm is the example I’m using here.

The third kind is work that people find interesting. For example, you discover a counterintuitive result or a new phenomenon, and then verify that it holds more generally. People find it interesting because they haven’t seen anything like it before.

The fourth kind is useful work. A lot of people want it done, but they don’t have the resources or the energy to do the hard work involved. When you do it, they think, “That’s exactly what I wanted.” A lot of benchmark work falls into this category. If you do it well, people find it useful.

Then there’s something I think you need to watch out for: research that doesn’t add anything to our understanding. You simply mix and match existing things, maybe improve performance by a couple of points, and people just say, “Okay, I know.” There’s no real new understanding; everyone already knew you could do that. Most capable researchers wouldn’t even bother doing that kind of work.

That’s the point I want to get across in this section on a paper’s value: when you publish papers, do valuable work.

Of course, I’m talking about your longer-term development here. Maybe you want an academic career, or maybe you don’t, but you do want to be a researcher. For a researcher, working on valuable problems is necessary. But if you’re just publishing a paper to help with graduate entrance exams or further study, then go ahead and get that paper published first.

Now let me be blunt about my own papers. Of my three papers at top conferences, the first was the USENIX Security one. I thought I’d proposed an interesting research question, and the reviewers recognized that. But judging it by my current taste, I think it was really just me entertaining myself.

The second was the NeurIPS paper. I think it did fill a small gap in the field, and the results were okay. But I’d say it barely qualifies as having a little bit of value.

The third is the S&P paper I’m currently submitting. I think it’s the most valuable work I’ve done so far, and the one I’m most satisfied with. I found something very important that others had overlooked, an aspect related to the “frontline,” and then told the community: we should be doing research at this level, and research at this level can actually be done.

So I think that work has more value. It’s the work I’m most satisfied with so far, and also the most solid work I’ve done.

How to Develop Research Taste

Let’s return to research taste. We’ve talked about why it matters and what it is. The third question is: how do you develop it?

I think developing research taste is fundamentally about developing intuition. Here’s a line from Einstein: “Imagination is more important than knowledge.” Physics is a subject that really requires accumulated intuition, and the very best physicists have extraordinary intuition.

Think about Einstein. Why didn’t he work on some other problem? Why did he study relativity? That’s his taste, his intuition.

For us, I think the first way is to train ourselves. This takes hard work: read lots of well-written papers and classic papers, and then ask why the authors chose that problem and where its value lies.

The vast majority of papers, essentially 99%, are gradually forgotten as time passes. How many papers from ten years ago do people still know today? Many no longer receive any attention. Only a very small number survive. Why do those papers survive? You need to think about that: why are they important, and why are the problems they studied important?

What did they mean for the field at the time? These are the questions you need to ask when you’re training yourself by reading classic papers.

The second way, I think, is through your environment, and this is the best return on your effort. Your taste will certainly be influenced by the people around you in your everyday life and work, by the taste of the people you interact with. When you talk to them often, their influence gradually rubs off on you, without you even noticing.

So I think one of the best things you can do for the effort involved is to put yourself in an environment where the people around you all have very good research taste. As their influence gradually rubs off on you, your own taste will improve, too.

That’s why the environment matters. I’ll come back to this later.

Building Your Presence and Influence in the Community

Besides taste, what else is there? I think the other half of being a researcher is building your presence and influence in the community. I’ll talk about this in three parts.

The first is to build your personal reputation. Do people know who you are? Can they remember you? You need to establish a recognizable identity in the community. You’re not just the author of a paper; you also need work that serves as your calling card. Your work can be better known than you are. People learn about the work first, then discover you and realize that you’re an expert in a particular niche.

The highest goal is that, when people mention a particular keyword, they think of your name. Or when they name a few people in a field, you’re among the first few they mention.

That’s your personal reputation. There’s also your academic credibility. First, there are some lines you cannot cross: academic integrity. You can’t submit the same manuscript to multiple venues at once, and you can’t fabricate results. Then, whether your code is usable and how well your experiments are done also help determine your reputation.

The second part is promoting your work, managing your visibility. You can use multiple channels. Outside China, people often use Twitter, now called X, and LinkedIn. In China, you might use Xiaohongshu, WeChat public accounts, or Zhihu columns. You can use these tools to let people know about your work.

Of course, personal blogs are used both in China and elsewhere. Then there’s what you release as open source. You can put your code on GitHub and your models on Hugging Face or ModelScope. Those can also serve as good calling cards.

Another important thing is giving talks. You might attend academic conferences or workshops and give presentations, or be invited to give a talk.

The third part is networking. You need to establish connections, socialize, and get to know more people, especially people in your research community. Talk to researchers in your immediate specialty and in the broader field. Networking is another way to increase your influence.

Let me add a few things on the side. If you want an academic career and a faculty position, you also need to think about funding and prestigious talent-program titles, what people in China call “hats.” Funding matters everywhere, but in China you also have to think about those titles. If you reach a certain age without one, your career has basically hit its ceiling. That’s a side point, outside the scope of today’s discussion.

Funding is interesting, too. I once talked about it with a professor over a meal. We discussed how funding differs between China and elsewhere. In China, most professors have government-funded grants and industry-funded projects.

With government-funded grants, for example from the National Natural Science Foundation of China, you have quite a bit of freedom in what work you do. But if you take on a project funded by a company, you basically have to work on what the company asks for. That’s how funding works in China.

But companies provide more money. So why might many professors take on a lot of industry-funded projects? Because those projects bring in more money.

Outside China, it’s actually the other way around: industry funding gives you more freedom. You can work on what you want, and the company doesn’t restrict your research direction, as long as you produce something. It doesn’t insist that you work on a particular topic. I find that interesting. There are some differences between China and elsewhere.

But one thing they have in common is that companies tend to provide more money.

Drawing on Institutional Prestige and Building Your Own Strength

Let me add one final point about the institution or research environment I mentioned earlier. It matters a lot, too. I think it essentially serves two purposes: drawing on its prestige, and building your own strength.

Drawing on its prestige means drawing on its reputation. This could be a university, an institution, a company, or even a particular lab or research group.

When do you need that advantage? When applying for recommendation-based graduate admission, which bypasses the entrance exam, studying abroad, or looking for a job. Any process that looks at your background and involves subjective assessment, such as an interview, will consider the institution you come from. A stronger institution gives you more prestige to draw on.

Different institutions give you different levels of advantage. That’s what people often mean when they talk about the importance of a title or affiliation.

The second part is building your own strength. What does that mean? It means using the resources your institution provides to develop your own abilities. What resources does it provide? There are mainly two kinds. One is people: those close to you, such as your fellow students and the professors who guide you.

Of course, the person who guides you is very important. Their ability can directly affect whether you get a paper published, so the quality of your guidance matters.

Then there are people farther away whom you can meet and get to know through the institution. The stronger the institution, the higher the level of people you can gain access to.

The second kind is material resources. A better institution can offer better equipment and instruments, or more GPUs and computing resources to support your experiments. That’s what I wanted to say about institutions and research environments.

Learning from Experience and Copying Someone Else’s Path

Finally, let me share a few personal thoughts. I know most of you are probably here because you’re aiming for recommendation-based graduate admission and want to hear about the experiences of students ahead of you. We were all the same. We looked at their experiences with graduate admission, research, and competitions, and then tried to follow their paths.

I think there’s nothing wrong with learning from their experiences early on, in your first or second year. But later, you need to step outside that frame. There’s something you need to understand: when you take them as role models and copy their paths, you can also end up limited by them. Their own perspectives are limited, too.

Standing here today, I don’t want you to copy everything I do either. There’s a saying: “Those who learn from me will thrive; those who imitate me will perish.” In my first year, I attended a session where students shared their experiences. Over these past few years, that’s the only session of that kind I’ve attended, the one in my first year.

There were four speakers, and only one had a paper, in a third-quartile SCI-indexed journal. At the time I thought, “Wow, that’s amazing!” My view is different now, but back then I really thought it was impressive. I didn’t know much, after all, and I thought a paper in a third-quartile SCI journal was already a big achievement.

Fortunately, my perspective opened up quite quickly after that, because I joined a group and started doing research. There were also senior undergraduates in the group. They used their past experience to judge what level of venue our project might reach. At the time, the students around me said we could publish at a CCF-B conference at best, one in the China Computer Federation’s B tier. I never imagined that we could publish a paper at one of the four top security conferences.

That’s a limit in your perspective, in what you know and what you can see. It’s understandable, though. After all, nobody around us had previously published at those conferences, let alone as an undergraduate first author. Even across the country, that was almost unheard of, so you can understand why they thought that way.

But it also shows that role models have their limitations. That’s the point I’m trying to make. And since I’m here today, with so many of you coming in person to listen, I should make it worth your while, right?

Examples of Undergraduates Doing Research

So let me introduce a few examples of undergraduates doing research. You can see how far an undergraduate can go when they reach a very high level. Of course, I’ve already finished the two main questions. What follows is a few more personal thoughts.

The first student has the strongest background, in terms of experience, of any undergraduate I’ve come across. During his undergraduate years, he studied in different universities and research environments and also did visiting research overseas. Those academic experiences were already outstanding.

He also has a lot of industry experience, having done research at companies both in China and abroad. His industry experience is excellent, too. The advisors he’s worked with are very impressive, and he’s had strong guidance in different research environments. His work has also accumulated a high citation count.

The second is another student who had overseas research experience as an undergraduate.

What’s particularly impressive about the second student? He has a first-author paper that has received a lot of citations on its own. And it’s not a survey or a benchmark paper; it’s an original piece of research. His overall citation count is high, too. With these two students, you can see that they have a lot of papers and a lot of citations.

Then there are two more examples, people who went straight into industry research after their undergraduate degrees. I learned about one of them at an academic conference. His first-author work had already achieved excellent results.

As an undergraduate, this student interned with the foundation-model team at a leading tech company. When he was about to graduate, the team recognized his ability and said, “Come join us.” So he did. He hasn’t been out of university for very long, but he’s already produced several high-level research contributions.

The other student is someone I really admire. He also went straight from his undergraduate degree to one of the world’s leading research organizations. I don’t actually know the person I learned about at the conference; I just know of him. The others I know personally.

The last student told me that AI is developing so quickly that you don’t know what the world will look like a year or two from now, or how far AI will have progressed. So he finished his undergraduate credits and degree early, and went straight into a research role at OpenAI after graduation.

That was something I’d never even heard of. Research positions at the world’s leading companies normally start with a PhD requirement. If you don’t have a PhD, they don’t recognize you as having research ability.

So getting into one of these top companies’ research roles straight after an undergraduate degree is very rare. And these people have already done excellent research. What I want to tell you through these two examples is that there really are many different paths in the world.

Look at them: they entered industry straight out of their undergraduate degrees and received offers for top industry research roles. For them, if they’re not interested in a faculty position, doing a PhD would simply be a waste of time. You still need a PhD for a faculty job, but if you weren’t planning to seek one in the first place, you’d be doing the PhD to get a better offer in industry afterwards.

These days, in industry, for example in China, the highest-paying offers are certainly research roles at major tech companies, such as AI research positions. The major companies have elite recruitment programs, such as Tencent’s Qingyun program, Alibaba’s Ali Star program, and Huawei’s Genius Youth program. These programs look at what you achieved during your PhD.

But for these people, if they can already enter those top companies’ research roles with an undergraduate degree, there’s no need to do a PhD. It would simply be a waste of time. And a year at a top research organization like that would teach you more than a year at any top university in the world. Your progress would certainly be faster, too. Okay, I’ve gone off on a tangent.

Finding Your Own Path and Having the Courage to Take It

That brings me to this: there are actually a lot of choices, a lot of paths in the world. But because our perspective is limited, we often see only a few: recommendation-based graduate admission, graduate entrance exams, studying abroad, getting a job, or taking civil service exams. When you think about where to go after graduation, what to do after your undergraduate degree, these are the usual paths that come to mind. But that’s also a limitation, a limitation in what we can see.

As I said earlier, there’s nothing wrong with following the paths of students ahead of you when you’re just starting out, in your first or second year.

But eventually, I think you need to find your own path. You need to think, explore, and figure out what you really want, and then decide which path to take on that basis. No two people have exactly the same experiences. You can find a path of your own.

The two people I just mentioned went straight from their undergraduate degrees into top companies’ research roles that would normally require a PhD. Those examples may feel too far removed from you. But there are examples closer to home as well. One student I know gave up the opportunity to continue his studies and chose to start a business. That kind of example exists, too.

Let me use myself as an example again. In my first and second years, my goal was probably much the same as that of the top-ranked students from earlier years in our department: I wanted recommendation-based admission to Tsinghua.

But in my third year, particularly over the winter break at the end of the first semester, around this time last year, I gradually changed my mind. I started to feel that Tsinghua wasn’t necessarily the best choice for me. In fact, even a senior student I knew at Tsinghua told me he didn’t think it was my best option. I hesitated for a long time, but in the end I chose another path.

What do I really want to say here? Making these choices outside the mainstream means taking on very significant risks. Those choices are not easy to make.

So the last lesson I want to share today is courage. I know some of you may be thinking: I’ll take a steady, conventional route, get into graduate school through recommendation, find a job, and do whatever I can. That’s okay. There’s nothing wrong with that.

But I also think there are some of you who want to try something different, who want to find your own path. Courage is the final lesson I want to leave with you.

When you feel trapped, when it seems there’s no way forward, courage will guide you to follow your convictions and find a path that is truly your own. Thank you, everyone.