Category: Personal Thoughts

A collection of my personal thoughts. Views are my own.

  • Meeting an Old Friend in Tech

    Meeting an Old Friend in Tech

    Last week, I caught up with a friend since middle school, who was visiting Philadelphia for a large language model (LLM) conference. He now works in tech, having been pursuing his M.S. and Ph.D. in natural language processing (NLP) and computer sciences at Carnegie Mellon and The University of Washington and is interning at Meta. It definitely felt a bit surreal to me that the last time we met face-to-face was in the winter of 2021, when I visited him and stayed at his place in Seattle, Washington. To compensate for this huge gap in time, I drove to Philadelphia twice over three days to spend a good chunk of time together and walked with him around Philly’s historic Center City.

    We had a number of topics to talk about. Having not been on top of the deep magics of LLMs, I did not dwell on the technical topics for too long. We chatted about our life, my new job, and how some of our mutual friends had been doing. He has proposed to his girlfriend earlier this year on a trip to Mexico but has not settled on a plan for the wedding, and she was busy participating in a series of architectural competition, doing her best to succeed at her architectural design firm in Los Angeles. A sentimental point came up when we realized it has been almost a decade since we left high school – that was when we took separate journeys and studied at different places in the US, and somehow, we were still in touch through all this.

    Getting close to graduation, my friend was actively job hunting, which quickly became a central topic of our conversation. Due to the slow market, he was feeling a lot of pressure and was yet to secure any offers. Among my friends, he has always stood out as one of the most innovative in the tech realm. He was also hardworking and mentally resilient, and I sincerely believed he would be a great engineer to advance the mission of any tech company on Earth (he also has >1,000 citations on Google Scholars to date), so I was quite empathetic with him about his pending offers that seem to be never coming.

    He admitted to wanting to work for the cutting-edge LLM companies like Open AI or xAI and pursue a career that is both lucrative and intellectually satisfying. We joked about how terrible a person Sam Altman and Elon Musk has turned out to be, including the board room intrigues Sam might have launched to push Open AI into the for-profit realm and the no-life approach of employee management Elon has turned to for the software company X and his AI venture. But wait a minute, did he not mention that he would like to join them? The image of him working crazy hours to experiment with their proprietary LLM models were as unreal as a full-scale AI-induced nuclear war. He had never struck me as someone who identified with the tech bro culture and his work ethics seemed to be at odds with that of those two companies.

    This perceived difference became even more obvious when we were chatting with another friend of his that went to the same conference about the extreme behaviors of people in the academia. I was shocked to hear a certain professor that I have heard of who sent a “?” to his undergraduate intern for not using his computing server during Christmas, and another world-renowned young assistant professor who did not let her top student to graduate in his 5th year, despite of his many achievements already. I could feel him and his friend detesting these behaviors. His friend even gingerly asked for tips for avoiding such professors in spite of their academic brilliance. I wondered, if these incidences sounded crazy enough, then how much of a difference can the companies like Open AI or xAI, both composed of geniuses that took work to an extreme level, make in terms of making their workplace humane. In light of this, my friend’s preferences still baffled me. Maybe my Epicurean believes prohibited me from taking a more reconciliatory look at the situations. Regardless, I could only wish him the best.

    On the flip side, a job in tech could easily bring my friend to places like New York City, where his salary and perks could make him live comfortably even in Manhattan. This was comforting news for me – it will be a fortunate way to re-unite with him after so many years and keep up with his ground-breaking works. Before we parted ways, I told him to keep me updated on his job status – despite his crazy dreams to chase the best of tech under strict bosses, it is always nice to have a long-term friend like him nearby.

  • A Few Lessons I Learned from My Ph.D. Program

    A Few Lessons I Learned from My Ph.D. Program

    A journey toward a Ph.D., as one may realize, could sometimes be harder than anticipated. There can be many ups and downs, some of which are intellectual and others emotional. There are also plenty of good things that can be found along the road. I have made it through my own journey by savoring my fair share of joy as well as having to deal with many cases of immense difficulties. In truth, besides finishing the dissertation satisfactorily, I realized that getting the most out of your Ph.D. experience requires a great deal of skills in self-management and relationship building, both of which allow you to cope with bad weather and enjoy beautiful sights. To help prepare you with the journey, I have written this blog to share a few but, in my opinion, crucial tips that I learned from mine in these regards. Hopefully, these tips can save you from headaches and self doubts when facing situations that are fraught with frustration and even dismay, and lead you toward a richer experience during your Ph.D. program by building a wonderful knowledge base for yourself, growing a network of collaborators, and being generous with sharing what you have learned on your very own path ahead.

    1. Own Your Projects
    2. Communicate with Your Advisors
    3. Manage Your Expectations
    4. Be Open-Minded
    5. Seek Collaboration
    6. Build A Sharing Mindset

    Own Your Projects

    When you enter your Ph.D. program, you might find yourself in uncharted waters of project management. First and foremost, there is often little to none formal training on some of the most important aspects, which include how to propose and begin a project, how to take a lead on project discussion, how to decide on actions post-discussion, and how to measure the overall progress. While you are uncertain of what is the correct course to navigate, many Ph.D. advisors tend to have the unspoken expectation that a Ph.D. student should be capably in charge of their work and manage their time accordingly. Furthermore, this expectation even applies when there are multiple projects, irrespective of their difficulties, that a student is working on simultaneously. Later, when you progress in your program and become more seasoned, it goes without saying that the pressure of leading and even owning your project only becomes higher.

    Such expectation can be justified as your advisor’s sincere hope for you to build up the habit of owning your projects, which is beneficial for you in the long run. Unfortunately, sometimes the advisor does not realize that this often is not emphasized enough to you or other students through clear verbal communication, which can result in more confusion or even self doubts. In my case, it took me more than a year since the beginning of my dissertation to figure this expectation out on my own. On a few occasions, my advisors would feel exceedingly worried about me when I did not push myself hard enough to make progress on my work or was waiting passively for their advice. Sometimes they would complain about this in a way that was borderline accusatory and hostile. Later on, once I took the lead in my projects and started to own them, they held back their complaints and acted much more understandingly and negotiable during our meetings.

    On the other hand, from the advisor’s point of view, they almost always have a handful of projects that they need to administrate to keep their results up-to-date and research funding coming, through which they could support students like you and I. As a result, their attention is scattered between all of these projects, which make them unable to keep up with the progress and nitty gritty of a specific student’s projects. In other words, they would default to relying on the student (like you and I) to remind them of a project details. When things get heated, this inability to keep up with the projects, coupled with their hefty expectation, could yield unpleasant engagements during their interactions with you and I. Even though their reactions are not always legitimate, it takes a student’s actions on owning their projects to better avoid such confrontations.

    So, what does owning a project look like in a Ph.D. program context? Let me first begin by clarifying that it does not involve getting to the ideal results quickly, as owing a project implies imposing some sort of control over it, and that the results of any research would not always be within your control. Instead, it might involve making strong efforts to clarify research questions when emailing your advisors, producing a clear project update or presentation during weekly meetings, or constantly contributing to the brainstorming sessions and technical discussion about your project/projects. These efforts not only show your decent grasps of the subject matters but also hint at your sense of organization and accountability to your advisors, which are precious traits in the inherently self-driven and disorganized academia. In addition, owing a project can also involve bringing a bright attitude or injecting positivity to the weekly meetings and project email chains. Eventually the good attitude will likely win over your advisor as well as your collaborators – who would not like someone that cheers everyone up when research itself is challenging enough?

    Communicate with Your Advisors

    It is fair to say that modern professional communication is expected to be smooth and unencumbered. To achieve this end, both sides of the communication should proactively check on the messages sent and received and get back to the other side timely and in an appropriate manner. Assuming that you keep up with these standards yourself, you would understandably expect your advisor to be on the same page. Unfortunately, from my own experience, some research advisors tend to largely live in their world and forget how to properly communicate with their students and peers, which often manifest itself through severe delays in email responses or unprofessional wording. Perhaps these are not all their fault, since there are usually no clearly written rules that govern the code of conduct for communication at research institutions. Needless to say, this also comes with consequences.

    At times, a lack of communication from busy advisors could signal a lack of attention and hurt a student’s feelings. At other times, a brusque and impatient email or message might leave a student on the receiving end confused about the advisor’s intention. I have heard certain advisors at other schools whom had said “this sucks/this is bad” to their students many times. If you have encounter some of them already, I feel you. In honesty, not much can be done in these scenarios other than maintaining a positive outlook yourself and not letting the lack of responses or demeaning words hurt your self worth. I personally think it helps to remember that in the vast majority of times, negative responses are not about you as a person but rather about the situations. It perhaps also helps to think that, after successfully dealing with the negligence/aggression, you will come out a stronger, more even keeled person.

    On the bright side, many research advisors and professionals I interacted with tend to be honest people, sometimes even brutally honest. If you could get past the “brutal” part of this, you could then take their honesty to your full advantage. This could mean that whenever you encounter some problems that needs their inputs, you could always raise the questions/issues and be forthright about what you want to know, no matter if such problems are technical or personal. Whatever issue you want to raise, it is important to do so early. Give your advisors a heads up plus a few days to let them reply will generally lead to more meaningful and constructive replies (and avoid unsavory comments in the meantime). You can always follow up if anything remains unclear. As discussed in the previous section, you should also NOT assume that your advisors are on top of your projects, so it is wise to break the questions down to small, self-contained portions, or try your best at giving a clear and concise description of the problems. Leading conversations is also a key part of owning your projects.

    As annoying as the communication styles of some advisors might be, they do not exonerate students like you and I from upholding our end of the bargain. Whatever were discussed previously also apply to ourselves when we craft emails or contribute to collaborative discussions. Think about what would happen if both sides decide to walk away from good communication practices. Most likely, it will be a downward spiral that never seems to stop.

    Lastly, do not be afraid of talking to your advisors about important milestones such as your graduation, job placement, or other topics about your personal development. After all, these milestones are a big part of what you sign up the Ph.D. program for. Again, be direct about them – some advisors sleep on these conversations or veer away from them (for one reason or another) unless they come from you in the most earnest form. Be ready to negotiate if your advisors are borderline unwilling to graduate you on time or talk about non-academic jobs. In such cases, leave email trails as evidence in case your advisors give you a hard time, and remember that, at the end of the day, it is up to you to get what you want – toughen up and do not be intimidated.

    Manage Your Expectations

    A subtle reader like you might have already deduced from the previous sections that having a healthy expectation of doing a Ph.D. yourself is almost as important as working hard. During the process of getting your Ph.D., not only might there be emotional ups and downs stemming from a lack of progress in your projects as well as bad beats in communication with your advisors, but doing research itself is almost like running a marathon, albeit with a moving finish line – you never really know when you are getting close unless you really are close.

    In my opinion, some active expectation management on your end is required to cope with this moving finish line, which includes being honest with yourself that you might face all sorts of hurdles at some point, while at the same time reminding yourself to not be bothered by what you cannot control. This probably sounds much easier said than done. In retrospect, I personally have perhaps spent way too much time worrying about having a bad day in research or a terrible interaction with my advisors. In the end, some of these worries had materialized as I imagined, but most did not. Had I followed my own advice better, I would have spared myself such internal struggles. If I could leave one further piece of advice here, it is that do not automatically assume a certain situation will end up in a way that blows up in your face. Also, as bad as these situations might be, they will not kill you, so there is no reason to fear them excessively.

    Meanwhile, you should strive for living a healthy and interesting life outside of your academic endeavors, which can serve both as an distraction when situations do not turn out the way you want and a grounding force to help you appreciate the beauty of life that also constantly exists around you. Maintaining some non-work-related hobbies, exercising often, and having a circle of people that support you (for example, mine consisted of my parents, my partner, and several close friends) will also give you a huge boost in your mental space and allow you to find peace within yourself, which in turn can often lead to healthier expectations towards your progress. In the end, you probably want to enjoy the positive aspects of your Ph.D. journey on your own terms and whenever you can, and keeping an interesting life outside of work might just help you achieve this.

    Be Open-Minded

    In previous sections, I listed a few pieces of advice to help you better cope with the bad and the ugly in your Ph.D. experience. From here on, I will transition toward writing about tips that can help you better enjoy the good of your Ph.D. life, starting with the first principle I employ to enhance my intellectual enjoyment – being open-minded.

    Unsurprisingly, various life gurus probably have already told you to be open-minded – about people, culture, art, food, so on and so forth. This lesson can be easily translated to the context of doing a Ph.D. and bring you a wealth of intellectual stimulation. Namely, it beckons you to not fixate your intellectual pursuit solely on a few topics highly similar or related to your dissertation. Now, don’t get me wrong – being laser focused on your own research areas and excelling at your dissertation work can definitely help you stand out, but with so much information being shared across the internet nowadays, research actually can be done differently from that of the last century. Bring on the interdisciplinary research.

    In my field (biostatistics), luckily, research is naturally interdisciplinary. I have witnessed people succeed at deriving the asymptotic bounds of risks and mathematically validating results from the cutting-edge machine learning applications, but I have also seen others enjoying a wide range of expertise – from genetics to imaging analysis, cancer to neuroscience, and surveying to sport analytics. Some people find themselves in the flow when they solve difficult math problems on their own, but others perform better when leading large collaborative projects. A few even pivot between different research domains. For example, undergraduate lab P.I. switched from leading a global consortium of imaging genetics to building *in vitro* stem cell models that could recapitulate key biology of early fetal brain development. One of my best friends during my Ph.D. program revolved his dissertation around three seemingly disconnected topics – visualizing cell-cycle in cancer cells that escape arrests after chemotherapy through dimensional reduction on multi-modal genetic data, developing a statistical test inspired by spherical geometry for cell-cycle differences, and designing the best when-to-treatment regimens for individual patients under resource constraints using reinforcement learning principles. He accomplished this through a blend of personal hard work and extensive collaborations between him and different labs or advisors, and his presentation of these topics was among some of the best I have seen despite his wide appetite research interest.

    If you restrict your research diet to one particular kind of domain, you will certainly not die of malnutrition. Nevertheless, you can still miss the opportunity to become a more well-rounded researcher. On the other hand, when you become more open-minded about other areas that are tangentially related to your research, you can better appreciate the efforts the entire research community put in into building the knowledge base for tackling complicated questions and even brainstorm crazy ideas with people from those other fields. If you are lucky enough, some of them could be the most interesting conversations you will ever have. I want to conclude this section with another piece of advice: sign up for one or two talks or conferences every year that are not the bread-and-butter ones people from your department regularly go. Those could be surprisingly fun!

    Seek Collaboration

    If you are someone who enjoys working on those group projects in college, a second way to elevate your Ph.D. experience is to seek collaboration with other researchers. This needs not be restricted to the collaboration recommended through your faculty advisors (although you should typically accept these recommendations, or else you might face a disappointed advisor when you are seeking other connections). If you can identify an avenue of collaboration through offering your unique skills set, you could simply reach out through an email on your own.

    There are multiple benefits to building collaboration like this. First, you could now embark on new research projects in a team environment. If you feel stuck, you could be well understood and helped by your teammates. If you make a breakthrough, you could easily break the news and feel the joy with the entire team. Simply put, collaboration is a good way to provide reliable relieves for the blues of solo research endeavors, like the ones for your dissertation. Second, if you intend to move deeper into research and are thinking about becoming a professor in the future, this could be your opportunity to build an early network of collaborators and mentors. Third, if you are lucky, the collaboration itself can lead to a co-authored paper, which could be added to your own portfolio. Last but not least, collaboration is the window to many exciting new areas that can broaden your research horizon. If you also remembered to keep an open mind about these, you will be immensely rewarded with the satisfaction of knowing more.

    In my own case, although my expertise in research was in genomics and high-dimensional data analysis, I branched out to collaborate with my friend who worked in a precision medicine lab. This move ended up giving me the opportunity to stumble upon and learn about all the cool concepts and applications of reinforcement learning, which is a tool that is popular in the precision medicine field but seldom mentioned in mine. I almost ended up doing a post-doc in that field too, if not that I decided I would be a better fit in the industry.

    Seeking academic collaboration of course entails being a good collaborator, which is something that you have to learn through experience but will be crucial to your long-term success in building your collaborative network. There are quite a few resources available that offer you some leads, including this one. Some of the earlier advice I posted, for example being mindful about communication, will also serve you well.

    Build A Sharing Mindset

    This is what I found out to be the most gratifying advice based on my own value – building a sharing mindset. I also wrote this very blog post because of this. Everyone knows that it takes a while to finish any research project in a meaningful way, but there is a wealth of intermediate knowledge that is generated through the investigations. Unless you are totally selfish, it would be a great pleasure to share the intermediate knowledge than to waste it, without compromising the originality of your own research at all.

    For example, in genomics and multi-omics, processing raw data from high-throughput sequencing is a pain in its own right. It requires many hours of computing on the cloud as well as many more of debugging the code and quality control of the outputs, not to mention there is not a very user-friendly way to do the whole processing. Unfortunately, other than a few long-winded websites, there is not much information on how to do this and can look very scary to the newcomers. However, once you have made it to the clean, processed data yourself, you could document the process as well as the code example and pass them on to the next batch of Ph.D. students undertaking this demanding task, while in the meantime answer any questions they might have. Not only will this boost the research outcomes of others, but you could also be credited in your lab or even in their works for helping them out, a win-win situation.

    Even after you finish your research, by walking an extra mile you could also leave something useful for other people. For example, you could build a free software package for the core algorithm that can be used in other applications, or you could present your new result at your school’s seminar or a local conference while waiting for the research paper to be accepted for publication at the journal you sent it to.

    As stated in this and my previous post, doing a Ph.D. is mostly a solo effort. When I encountered problems and needed to find some references or technical help, I did so mostly through many Google searches and clicks. Deep inside, I prayed that someone would have pointed me the way, no matter how tiny the pointer was, a Slack message or a quick link to the most related solution. Because of this solitude Ph.D. students like us face every day, any opportunity to share something you know with someone else in a similar boat will be doubly meaningful and appreciated. It is also why good instructors at this level are so renowned among their students. I sincerely encourage you to retain this sharing mindset throughout your journey and wish you can find joy sharing what you learned with others.

  • Transition to a Pharma Clinical Statistician – What I Have Learned So Far

    Transition to a Pharma Clinical Statistician – What I Have Learned So Far

    It has been nearly three weeks since I started with my new employer, Incyte Corporation in Wilmington, Delaware. I am assigned as a clinical statistician focusing on all-phase drug trials in dermatology, inflammation, and autoimmune diseases. Despite have a name that alludes to a manufacturer of biochemical weapons, Incyte has lived up to its true reputation—a mid-sized pharma company with good benefits and good people. From clinical scientists to chemists to my fellow colleagues in the biostatistics and programming department, everyone has left me with the sense that I can build strong working relationships and do good things, both for the company and its patients.

    Although Incyte’s firm stance against remote work means that I would come into the office four times a week, I’ve generally enjoyed the in-office days. In particular, I’ve taken advantage of the cafeteria’s discounted food options and seized opportunities to connect with colleagues through casual conversations and events like free ice cream socials.

    Reflecting on my first day, it’s clear that life in a corporation is a distant one from that in academia. Intellectually, gone are the cutting-edge research on new statistical methods and the mavericks exploring niche areas. In the world of pharmaceutical clinical development, what has worked before rules supreme, even if it’s something as straightforward as a two-sample proportion test, Pearson’s chi-square test, or ANCOVA. Virtually no one is interested in newly minted but lesser-known methodologies. On the upside, the head-scratching frustrations resulting from tackling difficult math problems, running complex numerical experiments, and diving deep into data solo have also faded. They’ve been replaced by a sense of collegiality. On my first day, our group director made it clear: if something doesn’t seem to work, the best solution is to ask for help rather than trying to figure it out on my own. The key is to get the deliverables ready on time—the practical stuff. Besides, no one wants to delay clinical studies. With dozens of trials at different stages running concurrently, teamwork is essential, involving clinical trial managers, database managers, clinical scientists, medical writers, statistical programmers, and us statisticians. When one person gets stuck, it affects others.

    Another lesson I’ve learned is how standardized everything is. There are set procedures for everything—from study design (including the protocol), database transfer, and analysis planning (centered around the famous statistical analysis plan, or SAP), to the actual analysis, reporting, and quality control. Clinical data is collected, stored, transformed, and analyzed according to the omnipresent CDISC (Clinical Data Interchange Standards Consortium) standard, which requires a deep dive into standard data domains, fields, and terminology. Additionally, most of the programming is done in SAS rather than R. Surprisingly, beyond contributing to protocol design and SAP writing, my main responsibilities as a clinical statistician mainly involve overseeing and quality-controlling various steps in the analysis pipeline and ensuring that everyone grasps the critical concepts and objectives. The bulk of the coding is handled by the statistical programmers based on the SAP specifics. As part of my training, I’ve been put in charge of a phase-2 trial that is about 50% complete, having been worked on by my predecessor, and I’m still learning my role within the ongoing process.

    A particularly positive aspect of my experience so far is how supportive everyone on my team has been and the efforts they have taken to make me feel comfortable. From the beginning, my manager and team director clearly communicated their expectations and priorities. The director even told me to be relaxed at work, noting, “If I give you something you can’t complete in time, it’s my failure as a manager.” They’ve also made a point of saying thanks to me whenever possible. During my PhD years, this kind of appreciation was something I seldom experienced from my advisors. At a team lunch yesterday, celebrating the FDA approval of our dermatological cream for treating pediatric atopic dermatitis (eczema)—a project I wasn’t involved in but was still invited to—I had a great time conversing with my manager about our backgrounds and families. Reflecting on my short time here, all these gestures have made me feel incredibly grateful for my co-workers and painted an optimistic picture for my early career.