Jinming Hu

Tue, 24 Mar 2026 12:00:00 +0800

I am now the founder and Chief Scientist of Sea-Land.ai.

I was a PhD student at the Department of Computer Science, University of Toronto. Before that, I graduated as a master student of artificial intelligence from the Zhejiang University CAD&CG National Key Lab ZJULearning Group. I was very fortunate to be advised by Prof. Deng Cai and Prof. Xiaofei He. I was also a research intern at the University of Toronto, where I was very unfortunate to work with Qizhen Zhang, so I decided to quit at last. My research includes machine learning, data mining, deep learning, computer vision, operating system, system programming, and database. I have worked as a system developer in Optiver Shanghai and have interned as a machine learning engineer in Hangzhou FABU, where I was fortunate to work with many colleagues, and in Google, where I was fortunate to work with Jingtao Wang. I was also a software engineer @ DolphinDB Inc, where I was fortunate to work with Davis and Xinjing Zhou, and many colleagues.

After My PhD Journey: My Conflict with My PhD Advisor Qizhen Zhang

Tue, 24 Mar 2026 12:00:00 +0800

When I applied for PhD programs, I received offers from several top database groups in Germany and Switzerland, as well as an offer from Qizhen Zhang at the University of Toronto. I eventually chose Toronto for database research. But after that, my visa to Canada was blocked for a long time and still not approved. So in September, I started doing research remotely. I had not formally enrolled yet, but it was still, in a loose sense, the beginning of my PhD journey. I wrote the post “Before My PhD Journey” at that time.

What I never expected then was that this journey would end so quickly. In less than three months, I decided to withdraw. And now, surprisingly, I am deeply grateful that I was able to leave so quickly.

Here is what happened. My first project was at the intersection of federated learning and databases. I chose this direction mainly because Qizhen had an important grant related to federated learning, and he wanted me to work on it. I was personally only moderately interested in federated learning, but I did want to work at the ML+DB intersection, so I accepted his suggestion.

Our first topic studied the impact of noise on federated learning. The paper was mostly experimental and found that federated learning is more sensitive to noise than traditional machine learning. During much of the project, we were like headless flies: besides me, there was no one in the group with strong machine learning background. So I was the one writing code, running experiments, and reasoning about the phenomena behind the results. We finished the paper in mid-October, submitted it to SIGMOD 2026, received a revision request in early January, passed the revision, and got acceptance in late February.

In parallel, because we observed the sensitivity of federated learning to data quality and noise, I wanted to solve that issue. I also noticed prior work showing that duplicate data does not help ML training much (especially for LLMs) and instead increases training time. Since LLM training data crawled from the web contains massive duplication, deduplication is highly needed. Existing methods were mostly centralized, so I proposed a distributed, decentralized, federated, privacy-preserving deduplication algorithm, so users’ private data would not need to be collected at a central node. I came up with this idea around mid-September, designed an efficient algorithm in about one day, and implemented it in about two days. We submitted it to WWW 2026 in early October.

After these two submissions in October, Qizhen pushed me to continue federated learning work, mainly on training framework and data pipeline optimization. By then, I was already less interested in continuing with federated learning. I could still do it well, but I felt it was not the most important near-term task. Privacy is important in the long run, yes, but right now the most pressing need is to improve intelligence and efficiency of LLM training and inference.

But Qizhen strongly wanted me to continue in federated learning. He told me: “During a PhD, what matters is deep cultivation in one direction. Getting distracted by other directions is harmful. I know your long-term goal is faculty; Toronto rejects many faculty candidates every year because their PhD research is not focused enough. You’re doing well in federated learning. I hope you can do even better and build your own brand.” I was puzzled. I had only been doing PhD-level research for one or two months. Yes, I had submitted two top-conference papers, and one later got into SIGMOD 2026, but that hardly means my direction should be fixed forever. Couldn’t I work on what interests me more? Later I realized the main reason was likely his grant needs.

Although I was uncomfortable, I kept working on federated learning. In late December, I traveled to Hong Kong with a friend to meet old friends, came back deeply moved, and wrote several posts. Then everything changed suddenly: one of my closest family members was diagnosed with a serious illness requiring long-term companionship and treatment. At that moment, I was shocked, but I immediately made the decision: I would not go abroad for the PhD. I would withdraw and stay with my family.

I informed Qizhen right away. He was also surprised, and said: “Don’t rush. You’ll soon have three top papers. Maybe I can contact the school and try to let you graduate in one year. You may not even need to spend much time in Toronto; maybe just come to handle formalities. I can’t guarantee it, but I can try.” Of course that sounded attractive: no need to go abroad, and still get a PhD in one year, which could keep my faculty dream alive. I did not expect he would go that far for me, and I felt both surprised and touched. I even told many friends about it. But my decision was already firm: no matter what, my priority had to be my family.

So what would I do after withdrawal? I could not bear seeing my family member suffer for years. I wanted to help overcome this difficult disease. That was unquestionably what I wanted most at the time. But I am not a medical researcher. What could I do? Then I thought: AI and LLMs are already very powerful. As a programmer and computer science researcher, I might feel that potential more directly than most people. I cannot personally cure disease, but if I can help push AI forward, maybe even help build a much stronger AI that can assist humans in conquering disease, then perhaps there is real hope. At that point I decided: I would start an AI company. That is why I chose AI entrepreneurship. One close friend strongly supported me and joined as a core co-founder.

An ironic part is that our team already had some public attention. We tried to stay low-key at first, but changes in company registration still exposed our startup. This triggered many self-media and marketing accounts to report on us, and many of those reports were likely AI-written and full of fabricated rumors, such as claiming I graduated from another school. It was frustrating. Up to now, we have barely spoken publicly, and we have never paid major domestic marketing accounts to publish PR pieces like “genius team charging at AGI.” Even the very few updates we posted about startup progress received heavy skepticism, especially on platforms like Xiaohongshu, where almost everyone assumed we were faking things. Still, one fortunate part is that many investors have tried to reach out and invest. We declined all of them. Right now we only want to focus on building.

In January, the SIGMOD revision request arrived. Qizhen started pushing me to revise. I spent time writing all required experimental code and designing experiments according to our hardware budget so all runs could finish before deadline. He remained very anxious and kept urging me. Even when I was busy with startup and family care and often got home after 11 p.m., he still pushed me for experiments, sometimes even asking for video calls to watch me work. As for the earlier promise to help me apply for accelerated graduation, he never mentioned it again. Still, out of responsibility, I completed all experiments on schedule. In the end, the SIGMOD 2026 result came out: the paper was accepted. The effort paid off.

Then came a dramatic turn. I logged into Slack to congratulate Qizhen, and found that in his lab group’s author list announcement, I had been moved to second author. I was speechless. On his personal website, the same paper also listed me second, with first author being an undergraduate who wrote zero lines of code and zero lines of the paper for this project. This was blatant first-author grabbing. I immediately messaged him on Slack: if you want to change author order, the minimum is to ask for my consent first. He then removed me from all lab groups, including the GitHub organization.

I complained to classmates and posted about it on WeChat Moments and Zhihu. I joked that I used to hear anti-PhD songs on Bilibili about first-author stealing, and never expected it to happen to me. Honestly I was not even angry. At the end of the day, it is one SIGMOD paper. If I can get one in a little over a month once, I can do it again. But I found it absurd: in prior conversations he was always approachable, and confidently promised to help me graduate in one year, and then eventually showed his real face.

A while later, I heard that an undergraduate in his group saw my Zhihu post and sent him a screenshot, so Qizhen responded by email. Ironically, the previous email right before that one had said: “I am very anxious that we may not be able to submit the revision. Please contact me as soon as possible. We should meet to decide what to do.” That was when there was still over a week before the deadline and he was anxiously pushing me to deliver all the experiments on time. I felt helpless then too: the jobs were already running on GPUs; pushing harder could not change physics.

In his latest reply, Qizhen said: when we submitted, I had told you that you and that undergraduate were co-first authors, and since his name comes earlier alphabetically, I put him first. I found this laughable. I wrote 100% of the code. He wrote 0%. I did all theoretical analysis and wrote most of the paper. He wrote 0 lines. On what basis were we co-first?

I immediately realized I had to defend ownership of my paper. So I wrote to the SIGMOD chairs right away to ask. The chair replied clearly: there is no such concept in the system here that allows arbitrary co-first author manipulation; I was the primary author, and no one could change author order without my consent. That made everything clear: this was Qizhen’s unilateral move. Maybe it was retaliation for my withdrawal and for the pressure he felt during revision. Looking back, the so-called accelerated graduation was probably just bait to keep me finishing the papers for him. Maybe even after finishing everything, first author still would not be mine. I later learned that his first PhD student had also master-out and left, and after that student decided to leave, he was similarly removed from the lab homepage; even that student’s once-first-author VLDB paper reportedly became an “N-th author” situation after revision. So I immediately posted my first-author paper on my personal homepage and open-sourced the code on GitHub. Otherwise who knows whether someday I would become “100th author.”

Looking back, life is full of randomness. You never know what the next period will bring. Someone who once seemed good may suddenly stab you in the back. A peaceful life may be disrupted by sudden disaster. What we can control is only our mindset and response. “A blessing in disguise” may sound cliché, but it rings true here. Leaving the PhD may look like a loss and perhaps the end of my faculty dream, yet entrepreneurship may still let me do what I most love: helping people grow. In a different way, it may also have helped me avoid a deeper pitfall. Maybe this is just life. What we can do is keep working, keep smiling at fate, never bow our heads, keep looking up at the stars, and keep our feet on the ground.

Before My PhD Journey

Tue, 24 Mar 2026 00:00:00 +0800

Sequel: After My PhD Journey

My PhD journey is about to begin. As I sit at my desk now, the familiar street outside my window looks the same, but inside me a long farewell has already started. Hangzhou in summer can be muggy, yet these days it has cooled down: still sunny, but no longer dizzyingly hot. It feels comfortable. More than five years of work pass before my eyes like a fast-forwarded film. I feel it is time to write down my experiences, thoughts, struggles, and choices. This is both an account of the past and a way to look toward the road ahead.

The idea of doing a PhD was never a sudden impulse. It was a seed planted in my student days. Back then, I was genuinely interested in research. I still remember a line from Andrej Karpathy’s “A Survival Guide to a PhD.” It felt like a beam of light. In a PhD, you may have five years to fully commit to one direction, become one of the people at the frontier, and push it forward with your own hands. That is a luxury, and perhaps a once-in-a-lifetime one.

“Luxury” is exactly the right word. In a world that is increasingly noisy and fragmented, having five uninterrupted years to focus deeply and purely on the boundary of a problem is itself a profound happiness. I longed for that luxury, for that feeling of moving through the wilderness of knowledge alone and with others, and for the excitement of intellectual collision with brilliant minds.

But practical realities always arrive before dreams. My family was financially poor. My parents had no pension and no medical insurance. They worked hard all their lives, and as their son I had the responsibility to provide at least a basic safety net for their old age. This responsibility was a non-negotiable constraint in my life. So between ideal and duty, I chose duty first. After graduation, I entered industry and folded the PhD dream away like a treasured white shirt, carefully placing it in the deepest corner of the closet. I told myself it was only temporary, and I would take it out when the time was right.

Searching in Reality: Finding My Objective Function

My first job was in high-frequency trading. By conventional standards, it was an excellent start: high technical bar, hard problems, and strong compensation. In the first months, I truly enjoyed solving technical challenges, and the intellectual satisfaction was real and intense. But over time, a deep sense of emptiness began to flood in.

I remember once working with a trader for two straight weeks on a core algorithm implementation. We were in the office almost every day until midnight. The whiteboard was packed with formulas and system diagrams, erased and redrawn again and again. In the end, we reduced strategy latency by 90%, a major technical breakthrough. The team celebrated briefly. But standing there, I felt strangely calm, even hollow.

Late that night, I walked out alone and stood on a bridge in Lujiazui. Lights glittered around me, traffic flowed below, and behind every window there seemed to be another busy soul like mine. I asked myself: what did all this change? Beyond making numbers move faster and capital games more extreme, did it make the world any better? I could not find an answer. It felt like forging an incredibly sharp weapon only to discover that its only purpose was to play a closed game irrelevant to most people. I kept asking: is this really the life I want?

This emptiness did not happen once. It seeped into my daily life. On morning commutes to Lujiazui, after late-night coding sessions, it showed up on schedule. The more successful the work, the stronger the feeling. I felt like a climber who reached a summit only to realize the view was never what he wanted. This mismatch between outward success and inner meaning was more exhausting than technical difficulty itself.

I began to realize I had become an efficient executor without a soul. My work did not resonate with my values. Painful as it was, this period was also a necessary calibration. It forced me to understand what I truly wanted: work that I personally find meaningful and that creates social value. I have a simple belief: if my work creates real value for society, then as a creator I will also receive corresponding returns, material or spiritual. If a job cannot satisfy my pursuit of meaning, then no matter how high the income, it becomes long-term internal friction.

So I described my life confusion with engineering language: my life’s objective function was wrong. No amount of local optimization can fix a mis-specified goal. I did not need to run faster on the wrong road. I needed to stop and find the road that truly belonged to me.

At that moment of confusion, a light came from afar. About a year after graduation, my close friend Xinjing, who had once discussed technology and dreams with me at Zhejiang University, was admitted to MIT for a PhD to pursue databases. I was sincerely happy for him, and I also felt a sharp, almost painful envy. His choice rang like a clear bell in my heart, reminding me that the white shirt in the closet was still clean and new. I told myself: I must walk that road too.

An Unexpected Rehearsal: Finding My Calling on the Podium

After leaving my first job, I got an accidental chance to return to Zhejiang University and teach machine learning for the Turing class for a period of time. At first I only wanted to fulfill my advisor’s request and fill a temporary gap. I never expected this “accident” to become a decisive turning point in my life.

That period remains unforgettable and exciting to this day. The first time I stood on the podium and saw dozens of clear, bright, eager eyes, I was struck by a completely new feeling. I was no longer a confused engineer. I became a transmitter of knowledge and a guide for thought. At that moment, it felt like the white shirt I had folded away floated back out of the closet and gently settled on my shoulders, carrying the smell of sunshine.

I spent enormous time preparing lectures. I almost gave up all social life, and immersed myself in syllabi and lectures from MIT, Stanford, CMU, and Berkeley. I wanted to offer these brilliant students a genuinely world-class course. In class, I did not just teach formulas and algorithms. I wanted to convey the joy of learning and the motivations behind ideas. I guided students to ask questions and encouraged critical thinking.

I remember once, while discussing spectral clustering, a student raised a hand and asked: “The min-cut problem seems very close to max-flow min-cut from algorithms. Can we just use that algorithm directly?” I paused for a few seconds and then felt a surge of joy. This was exactly what I most wanted to see: students actively building their own cross-disciplinary understanding, instead of passively receiving knowledge. Seeing confusion become clarity through explanation and dialogue brought me a deeper and more lasting joy than writing any efficient code ever had.

In the final class, I told them that I hoped this course had shown them that machine learning is both useful and interesting, and that learning can also be meaningful and joyful. I recited a passage from “Young China” and told them: in the small sense, many problems in machine learning and deep learning remain unsolved; in the large sense, this country and this society still have many unresolved problems. I believed some of them would solve part of those problems in the future, and create their own future. At the end, they gave me a long applause and came up for photos. Looking at their young and hopeful faces, I felt absolutely certain: this is what I want to do for life.

That experience also made me reflect on my own path. My family was poor. Without this country’s educational policies, I would never have had the chance to come this far or to improve my family’s life. Education is the most precious gift this country gave me. Now I want to pass that gift on. I resolved to contribute to computer science education in China, and to students from ordinary backgrounds who still carry great dreams, just like I once did.

At that moment, the PhD dream in the closet perfectly overlapped with a mission of serving education. It was no longer only a personal academic pursuit. It gained a larger and heavier meaning.

Tempered in Engineering: Forging Weapons for the Ideal

I knew this was what I wanted to do for life. But the path would be long. Ideals need the soil of reality. That white shirt representing the future had to be redeemed through concrete effort. I still needed to build financial security for my family, and I also needed to sharpen stronger tools for future academic work. With this renewed clarity, I joined DolphinDB. The following years became a period of rapid growth in my engineering capability.

I was very lucky to work in a technically pure and highly challenging environment, and to experience the full process of building a world-class software product from scratch. This was not a school assignment, nor tightening screws in a giant company. We faced real-world extreme performance requirements and complex scenarios. Those years were intense tempering, and what I learned went far beyond writing code.

In 2021, we started from zero to develop a new storage engine. Everything had to be built from scratch. Xinjing and I did extensive design research, read top papers from the past decade one by one, and debated repeatedly in front of a whiteboard, drawing version after version of architecture diagrams and stress-testing feasibility in theory. I still remember discussing one key memtable design for days: variants of LSM-Tree, applicability of B+Tree, and our own hybrid proposal. Every option was decomposed down to details, with prototypes and experiments used repeatedly to validate ideas.

Perfect plans in theory always meet surprises in practice. We had to write minimal prototypes for proof of concept. I still remember when we first implemented lock-free and zero-copy memtable and saw system throughput jump by an order of magnitude in testing. That pure engineer’s joy is hard to describe: turning theory into reality, and turning insight into performance.

I also started taking on the role of R&D lead, organizing team work and task assignment. More importantly, I stayed on the front line with everyone and kept coding. I believe only by going into implementation details can one truly understand a system’s strengths and weaknesses. We set extremely high quality standards, wrote large-scale tests, and built solid CI/CD pipelines to ensure every release was stable and reliable.

That full loop transformed my understanding of how to do engineering well. I am deeply grateful to my friend Xinjing, who introduced me and became my first mentor; to Davis, founder of DolphinDB, who is not only an outstanding CEO but also a top engineer still coding on the front line; and to Da Fei and every teammate who helped and guided me.

At DolphinDB, my work was recognized quickly and I became R&D lead in under four months. I could have stayed, waited for IPO, and obtained substantial financial rewards. During those days, I often felt a happy contradiction: on one hand, I enjoyed the pure joy of solving world-class engineering problems with excellent teammates; on the other hand, late at night, the classroom lights and students’ eyes would return vividly in my mind. That calling kept reminding me: all the tools I was forging here were for a more important battle in the future. The white shirt was still in the closet, waiting to be worn.

In fact, before joining, I had been open with Davis about my long-term plan. I told him I would definitely pursue a PhD in the future, and that one key reason for working here was to build financial security for my family first. He understood and respected that choice. Over these years, I never stopped preparing for PhD applications. I was fortunate to collaborate with Xinjing, Viktor Leis, Xiangyao Yu, and Mike Stonebraker, and publish a SIGMOD 2025 paper, which built a solid foundation for my application.

Decision Time: When Dreams Meet Reality

Time passed, and I reached my fourth year at DolphinDB. I calculated my savings and felt they were enough to provide my parents with a reassuring baseline for the future. The constraint that had once bound me was finally satisfied.

I knew it was time to open the closet and take out that long-preserved white shirt.

I began preparing PhD applications formally. Because of my undergraduate background (Northwestern Polytechnical University), US visa policy posed substantial uncertainty, so I had to focus on other countries. During this process, I once again felt the power of kindness and support. Viktor Leis, second author of our SIGMOD paper and professor at TUM, kindly introduced me to several top professors in Germany and Switzerland. Thanks to this, I entered their view and ultimately received offers. I remain deeply grateful for that cross-border support.

At first, Canada was not on my list. But my master’s advisors, Prof. Deng Cai and Prof. Xiaofei He, strongly encouraged me to consider North America. They believed it remained the center of computer science research, with the most active academic ecosystem. My mentor from my Google internship, Jingtao, also strongly recommended that I apply and gave me many practical suggestions. Encouraged by them, I submitted an application to the University of Toronto.

When all offers arrived, I had the pleasant difficulty of choosing. After in-depth conversations with several respected mentors, they almost unanimously recommended Toronto. During my PhD, I also wanted to explore innovation at the intersection of databases and AI, and Toronto unquestionably has world-class resources in that area. So I made my final choice. Later I reflected that where I do research may not matter that much in itself. Those schools all provide excellent environments. For me, being able to quietly do what I love and move toward my goal is enough.

When I told Davis I had decided to pursue the PhD, he was reluctant to lose me. But he told me something deeply moving: “Of course we hope you stay and go all the way to IPO with us. But your own path matters more, and we support your choice.”

Some Reflections Before Departure

Looking back at these five-plus winding years, I want to share a few reflections.

First, choosing what you truly love is profoundly important. For me, pursuing a PhD and taking the education path feels like an irresistible calling. I can only choose this; otherwise I would regret it for life. Some life choices are beyond cost-benefit calculation. When you hear a strong enough inner voice, you only need to listen and follow. Because of that, I believe this love can sustain me through whatever difficulties lie ahead. As Jobs said, “everything else is secondary.”

Second, in the AI era, if we use tools well, individual capability can be amplified in unprecedented ways. I am optimistic about the future. I believe these technologies can help create education models that are more efficient, fairer, and more inspiring. I hope to combine what I learned in academia and industry with AI to make practical contributions, however small, to computer science education in China.

Finally, I want to thank everyone who helped me along this path: my advisors, who gave both knowledge and direction at key moments; professors who wrote recommendation letters and trusted me; my friend Xinjing, both role model and fellow traveler; like-minded friends whose conversations constantly sharpened my thinking; my family, who silently bore my “stubbornness” and gave unconditional love; and the leaders and colleagues at DolphinDB, whose support and mentorship are invaluable assets for my future journey.

And the person I must thank most is my wife. When I described this difficult and unusual path and asked for her support, she gave me unwavering understanding and strength, allowing me to pursue my calling without looking back. I still remember telling her my favorite declaration of love was Premier Zhou Enlai’s line to his wife: “I hope we can devote ourselves to the revolution together, and one day walk to the guillotine together.” She understood instantly and held my hand tightly. She is my comrade on this journey.

My country, I will come back. Wait for me.

Mon, 01 Jan 0001 00:00:00 +0000

I am now the founder and Chief Scientist of Sea Land.

I graduated as a master student of artificial intelligence from the Zhejiang University CAD&CG National Key Lab ZJULearning Group. I was very fortunate to be advised by Prof. Deng Cai. My research includes machine learning, data mining, deep learning, computer vision, operating system, system programming, and database. I have worked as a system developer in Optiver Shanghai and have interned as a machine learning engineer in Hangzhou FABU and Google. I was also an software engineer @ DolphinDB Inc.