Conversation with Prof. Owolabi Legunsen on Research Fit, Advisor Capacity, and Keeping Doors Open

Introduction

I recently had a thoughtful and generous conversation with Prof. Owolabi Legunsen about research fit, advisor switching, and how to think about an uncertain first year in a Ph.D. program.

The conversation was especially helpful because he was both candid and supportive. He made two things clear at the same time:

• I should take my current situation seriously and think carefully about research fit.
• I should not interpret early uncertainty as evidence that I am failing as a Ph.D. student.

His advice was grounded, realistic, and humane. The discussion ranged from my difficulty aligning with my current group’s newer hardware-oriented direction, to possible faculty I could speak with, to the more general point that finding the right problem often takes iteration.

The Core Issue: Research Fit

I explained that when I applied, I understood Prof. Zhiru Zhang’s work as being closer to areas like compilers, architecture, and code analysis for hardware. More recently, however, much of the work I had been exposed to seemed to move more toward hardware design and verification.

That shift matters because my own background is still much more rooted in software:

• software engineering,
• programming languages,
• program analysis,
• and related tooling-oriented thinking.

As a result, I felt that I was spending a large amount of effort simply trying to build enough background to understand the new space, before I could even identify a concrete research problem.

Prof. Legunsen’s response was not to immediately conclude that this meant there was no future there. Instead, he framed the situation as involving two very different possibilities:

1. there may truly be no meaningful space for my background in the direction my current advisor wants to pursue, or
2. there may still be room for me, but I may need to work harder to identify how my skills could fit into that agenda.

This distinction was one of the most useful parts of the conversation. It pushed me away from a vague feeling of mismatch and toward a more precise question: is the problem a fundamental lack of fit, or is it an unresolved fit that still requires more iteration?

His First Principle: Make Sure the Door Is Really Closed Before Walking Away

One theme came up repeatedly: before deciding that working with my current advisor is no longer viable, I should be able to say honestly that I did everything in my power to make it work.

Prof. Legunsen said that it is actually uncommon for there to be absolutely no future between a student and an established advisor. His intuition was that if a professor has already invested time in thinking about how a student might fit, then the first step should be to keep exploring that possibility rather than abandoning it too quickly.

That did not mean ignoring genuine misalignment. Rather, it meant being disciplined in how I evaluate it.

He wanted to understand what topics had already been explored with Prof. Zhang. I described several examples:

• an early idea around operator fusion in machine learning pipelines,
• a possible effort involving AWS Neuron and low-level accelerator programming,
• and later discussion around a grid-based accelerator-design methodology and related hardware-design questions.

A recurring issue across these possibilities was that I often felt I was being asked to operate in an area where I lacked enough prior context to propose strong problems quickly. For example, when I tried to make progress by exploring infrastructure such as CIRCT, that kind of progress did not seem to count as much as identifying the right research problem up front.

From Prof. Legunsen’s perspective, though, the bigger question was still whether there remained some way to connect my strengths to Prof. Zhang’s broader agenda.

Flexibility and Iteration Matter More Than a Perfect Initial Match

Prof. Legunsen gave what was probably the central strategic advice of the conversation: the key words for me should be flexibility and iteration.

His view was that I may need to reimagine what kind of project I could do within my current advisor’s group, rather than hoping that the lab’s direction will become naturally closer to my preexisting interests.

That point was direct but important. In effect, he was saying:

• an advisor does not necessarily need to reshape a research agenda around a first-year student,
• so the student’s task is often to search for a problem that fits the advisor’s current momentum,
• and the first few attempts at finding that problem may fail.

He emphasized that this is normal. Even for students who are doing fine, the first problem is often not the right one.

I found this framing reassuring because it replaces self-judgment with process:

• talk,
• try something,
• learn from the mismatch,
• iterate,
• and keep the conversation going.

Health Comes First

I also shared that I had been dealing with mental health difficulties and that I had taken an approved health leave of absence for the semester.

Prof. Legunsen responded with a level of care that I appreciated very much. He said clearly that health comes first and that taking care of myself should be the priority. He also asked whether I felt I was getting the support I needed and whether things were improving.

This was not a minor side note in the conversation. It shaped his overall advice. Since I am on leave and not returning until next semester, he thought it was reasonable not to force immediate decisions right now.

That perspective matched what I had also heard from the DGS in ECE: while on leave, I should focus on recovery, and then continue exploring research options when I return.

A Candid Answer About Advisor Capacity

Another valuable part of the conversation was that Prof. Legunsen was completely unambiguous about his own advising capacity.

He told me that his group currently has six Ph.D. students, while his normal capacity is four, so he is already oversubscribed. Because of that, he is not looking for new students, and switching to his group is not a realistic option right now.

I appreciated this clarity. It prevented the conversation from becoming vague or performatively encouraging. He was willing to help me think, but he did not want to suggest a possibility that he could not responsibly support.

He also explained that taking on another student in that situation would be unfair both to that student and to the students he is already advising.

That was a good reminder that advising decisions are not only intellectual; they are also constrained by time, energy, attention, and funding.

Faculty He Suggested I Contact

Because my background is still closer to software than to hardware design, I asked which other faculty at Cornell might be worth contacting.

Prof. Legunsen sketched the landscape in a helpful way.

In software engineering and programming languages

He suggested that I continue following up with or reaching out to:

• Saikat on the software engineering side,
• Alexandra Silva on the programming languages side,
• Andrew Myers,
• and Justin Xu.

He also noted that some faculty may be on leave or slow to respond, so lack of a quick reply should not be overinterpreted. His advice there was simple and practical: send reminders and do not take delayed email personally.

In systems and adjacent areas

He also mentioned a few people outside the most obvious software-engineering / PL buckets:

• Emanuel Trummer, whose work is more database-oriented but increasingly intersects with large language models,
• and Sanyam Kapoor, also in databases, but with broad familiarity with software-engineering-style questions.

At the same time, he warned that faculty working more deeply on hardware-level concerns such as networking, caches, throughput, and closely related architectural topics might not actually reduce the ramp-up burden I am currently trying to manage.

In ECE

He also pointed out that it was interesting that most of the people I had contacted were in CS rather than ECE.

Within ECE, he suggested that I might eventually speak with:

• Christopher Batten,
• and, to a lesser extent, Jose Martinez.

His reasoning was that they may have some overlap with the parts of Prof. Zhang’s world that originally attracted me, even if they are not exact matches to my previous software-oriented background.

His Advice About Prof. Zhang Specifically

Although he suggested multiple people to contact, Prof. Legunsen still repeatedly returned to the possibility that Prof. Zhang may remain the most realistic long-term fit.

His reasoning was straightforward:

• Prof. Zhang is the faculty member who has spent the most time trying to think about how I might fit.
• I am still in my first year.
• My current difficulty may reflect an incomplete alignment process rather than a final negative verdict.

He encouraged me not to close that door.

One especially concrete part of his advice was to use my upcoming Amazon internship, which will involve compilers for the Neuron platform, as an opportunity to notice technical problems that might connect back to Prof. Zhang’s interests.

His suggestion was that I should:

• pay attention to concrete problems during the internship,
• keep Prof. Zhang in the loop,
• and run possible research ideas by him as they emerge.

This was a useful reframing. Instead of treating the internship as separate from the advisor-fit problem, he saw it as a possible source of evidence and ideas.

A Reassuring View of Early-Stage Ph.D. Uncertainty

Near the end of the conversation, Prof. Legunsen said something I found especially encouraging: I should not tell myself that I have been a disaster or that something is wrong with me just because I have not yet found the right project.

He emphasized the phrase “so far”.

Not having found the right research problem so far is very different from not being capable of doing so.

He even shared that when he started his own Ph.D., he also struggled to find a project, and his first paper did not come until his third year. That context mattered. It made the difficulty feel less like personal failure and more like part of the normal uncertainty of research training.

His broader message was that:

• the Ph.D. is not won in the first year,
• fit often takes time,
• and keeping doors open is usually wiser than making dramatic conclusions too early.

Overall Takeaway

My main takeaway from this conversation is that Prof. Legunsen offered a combination of realism and encouragement that I found deeply helpful.

The realistic part was:

• he cannot take new students,
• I should not assume that moving to another advisor will be easy,
• and I may still need to adapt significantly if I want to succeed in a hardware-adjacent environment.

The encouraging part was:

• I am not obviously in academic trouble,
• early difficulty finding a research fit is not unusual,
• my health should come first,
• and I should keep iterating rather than concluding too quickly that everything has failed.

If I had to condense his advice into a few phrases, they would be:

• take care of your health first,
• do not close doors prematurely,
• be flexible,
• keep iterating,
• and do not confuse “not yet” with “never.”

I left the conversation with a clearer picture of what to do next: recover, keep talking to people, continue communicating with Prof. Zhang, follow up with other faculty where appropriate, and use the summer internship as a chance to discover more concrete technical directions.

Overall, I am very grateful for Prof. Legunsen’s honesty, kindness, and perspective.