Guide · Research
Most students never ask, and the ones who do usually ask badly. The process is unglamorous: read what a professor published, email them about it specifically, and be useful on a small task before you expect anything larger.
Research is one of the few things an undergraduate can do that is genuinely rare among applicants. Coursework is common, projects are common, internships are competitive but common. A publication with your name on it is not common, and it signals something specific: that you can work on an open problem for months without a solution key and write it up to a standard that survived peer review.
It also compounds. One paper makes the second easier, gives you a reference who has seen you work rather than seen your grades, and opens graduate admissions if that path ever becomes interesting.
"Undergraduate research" covers at least five arrangements with different costs, different commitments, and different eligibility rules. Knowing which one you are asking for makes the email clearer and stops you asking for something the professor cannot give.
| Arrangement | What you get | Typical commitment | What it costs you |
|---|---|---|---|
| Volunteer | Access, mentorship, a reference, and the fastest route in | 5–10 hours a week | Unpaid, and it competes directly with a paid job |
| Course credit / independent study | Units toward the degree, a grade, a defined deliverable | Often around 3 hours a week per unit | Tuition per unit, and a graded deliverable on a fixed date |
| Hourly or funded assistantship | Pay, and usually a scoped project | 10–20 hours a week | Frequently depends on work authorisation and on grant timing |
| Structured summer programme | A stipend, a cohort, and a poster at the end | 8–10 weeks, full time | Applications close months ahead; eligibility rules are often strict |
| Supervised independent project | Maximum autonomy, a possible first-author paper | Whatever you make it | Almost no structure, so it fails quietly if you do not self-manage |
Two practical notes. Volunteering for a term is the shortest path in and it is the one most people can only afford briefly, so treat it as an entry ramp rather than a plan — say up front that you are hoping to move to credit or funding next term and most supervisors will help arrange it. And funded programmes frequently carry eligibility conditions on citizenship, residency, or work authorisation, which is the first line to read rather than the last; the funding guide covers how those clauses are usually worded and where they break.
You are not looking for "a research opportunity". You are looking for a specific person whose specific work you can describe in a sentence. That distinction is what separates an email that gets a reply from one that does not.
Do this in one two-hour sitting with a spreadsheet open, not across three weeks of good intentions. One row per person, and the row is not finished until every column has something in it — the empty columns are exactly the ones the email needs.
name | department | one-line research area, in your own words | most recent paper title (exact) | year | venue | one specific thing in it you can comment on | skill of yours it plausibly needs | lab site? open positions listed? application process? | who could introduce you | emailed on | followed up on | outcome Stop when you have 8 complete rows. Do not email anyone whose "one specific thing" column is still blank - that blank is the reason generic emails read as generic.The unfair advantage for a CS undergraduate is that most labs outside CS have data they cannot process well. Being the person who can write the pipeline, clean the dataset, or build the study tool is a legitimate entry ticket even before you understand the domain.
You cannot read five papers properly, and you do not need to. The common approach is three passes at increasing depth, and for a cold email you need the first pass on five papers and the second pass on exactly one.
| Pass | What you read | Time | What you should be able to say afterwards |
|---|---|---|---|
| 1 | Title, abstract, introduction, section headings, conclusion, a glance at the references | 5–10 min | What they claim, roughly how they got there, and whether it interests you |
| 2 | Figures, tables, and the method at a level you could describe. Skip proofs and derivations | 45–60 min | How the study was designed, what was measured, and where you find it unconvincing |
| 3 | Reconstruct the work — re-derive, re-implement, or re-analyse | 4+ hours | Whether it is correct, and what you would do differently |
Pass two on one paper is what generates the sentence that gets you a reply. Not a compliment — an observation. A limitation the authors named in their own future-work section, a measure you would have expected and did not find, a population you wondered about. Future-work sections are the most under-read paragraphs in academia and they are a list of things the author already wants done.
A professor receives a lot of generic requests and ignores nearly all of them. The three things that get a reply are: evidence you read their work, a concrete offer of what you can do, and a small ask. Keep it under 200 words.
Subject: Undergraduate interested in [specific topic] — [your name] Dear Professor [Name], I am a [year] [major] student at [university]. I read your paper "[exact title]" and found [one specific thing — a method, a result, a limitation you noticed] genuinely interesting because [one honest sentence, in your own words]. I have [concrete relevant skill: Python, data cleaning, building study tools, statistics] and used it to [one line about a real project]. I would like to contribute to your lab, including on the unglamorous parts — data collection, cleaning, transcription, tooling. Would you have 15 minutes in the next two weeks to talk about whether there is anything I could help with? I have attached my CV. Thank you for your time, [Name] · [email] · [portfolio or GitHub link]Shorter, because the hard part is already done. The one thing to include is a reminder of which version of you they are being asked to remember.
Subject: [Course code] student, interested in your [topic] work Dear Professor [Name], I was in your [course, term] - I sat near the front and did the [specific project or assignment]. I have since read "[exact title]" and [one specific observation]. I can [skill], and I would be glad to start on data cleaning, literature screening, or tooling. Do you have 15 minutes in the next two weeks? [Name] · [email] · [link]Here you are being screened against stated criteria, so answer them in the order the posting listed them. Do not make the reader map your paragraph onto their requirements.
Subject: Application - [exact position title as posted] Dear [Name / Lab], Applying for [exact title], posted [where]. · [Their requirement 1]: [your one-line evidence, with a link] · [Their requirement 2]: [your one-line evidence] · [Their requirement 3]: [honest answer, including "not yet, but"] Availability: [hours per week], [start date], [through when]. CV attached. Portfolio: [link]. Thank you, [Name] · [email]One follow-up, ten days later, replying to your own thread so the original is attached. Non-reply is inbox volume in the overwhelming majority of cases, and a second email is not an imposition — a third one is.
Subject: Re: [original subject] Dear Professor [Name], Following up on the note below in case it arrived at a busy time. Still very interested, and still happy to start on the unglamorous parts. If the lab is not taking undergraduates this term, I would be grateful to know that too - and if there is someone else you would suggest I contact, I would appreciate the name. Thank you, [Name]It is almost never the interesting part, and this is the point at which most undergraduates quietly disappear. Expect literature screening, data cleaning, transcription, annotating responses against a codebook, running participant sessions, or building a small internal tool nobody outside the lab will ever see.
Do it well and finish it on time. The entire progression from data-cleaner to co-author runs on that one behaviour, because a lab's scarcest resource is a person who reliably completes things without supervision. The students who ask for a research question in month one and vanish by month three are the norm; being the exception is not difficult, only consistent.
An Institutional Review Board is the university committee that reviews research involving human participants before it starts. If your study involves people — surveys, interviews, classroom studies, usability sessions — you need approval, and running the study first is not recoverable: data collected without approval generally cannot be used or published.
In practice it means a written protocol describing who the participants are, what you will ask them, what data you collect, how it is stored, how it is anonymised, and what the consent form says. As an undergraduate you will typically complete a short human-subjects training course and be added to an existing protocol rather than writing your own.
| Stage | Who does it | Realistic duration |
|---|---|---|
| Human-subjects training certificate | You, once, valid for a period of years | 4–8 hours |
| Protocol drafting | Usually the PI, with your sections | 1–3 weeks |
| Internal or departmental read-through before submission | Lab or department | A few days to 2 weeks |
| Submission and category determination | The board | Days |
| Review, for lower-risk categories | The board or a delegated reviewer | Commonly 2–6 weeks |
| Review requiring a full convened board | The board, at a scheduled meeting | Tied to the meeting calendar, often monthly |
| Revision round — assume at least one | You and the PI | 1–3 weeks |
| Approval to first participant | You | Recruitment can start immediately |
| Any later change to the protocol | Amendment, reviewed again | Weeks, again |
Add that up and the front of a human-participants project is realistically six to twelve weeks before the first participant. Timings vary widely by institution and by risk category, so ask the lab's coordinator for their actual numbers rather than trusting these. The planning consequence is the same either way: a study you want to run in April needs its protocol moving in January, and "we will sort out the IRB later" is how a term of work becomes unpublishable.
Mixed-methods research combines quantitative data — scores, timings, counts, survey scales you can run statistics on — with qualitative data — interviews, open responses, observation notes that you analyse by coding for themes.
The reason to combine them is that each covers the other's blind spot. Quantitative data tells you that something changed and how much; qualitative data tells you why, and often surfaces the explanation you would never have thought to measure. A typical design runs a measurable intervention, collects the numbers, then interviews a subset of participants and codes the transcripts for recurring themes that explain the numbers.
Undergraduate authorship is rare enough that it is memorable in almost any application pile. Beyond the signal, the process teaches things coursework does not: how to scope a claim so the data actually supports it, how to write for reviewers who will attack the weakest sentence, and how to respond to criticism without either collapsing or arguing.
Realistic routes in, roughly in order of accessibility: a poster at a conference or a departmental symposium, a workshop paper, a short paper or experience report, then a full paper as a middle author. Posters are not a consolation prize — they are the normal entry point, and they still involve peer review, a deadline, and defending the work in person.
Conventions differ by field and getting this wrong causes more bad feeling in labs than almost anything else. In most of computer science, author order reflects contribution: first author did the most and usually wrote the draft, and the last position is commonly the supervising principal investigator. In theory, mathematics, and economics, alphabetical order is standard and carries no ranking information at all — which is worth knowing before you read anything into a name's position.
Research is worth doing regardless, but it only counts in an application if it is legible to someone who was not there. Three surfaces, three different jobs:
Keep a running log from week one — what you did, dates, numbers, tools. The resume line then writes itself in ten minutes rather than being reconstructed badly from memory during an application deadline.
My own path is not exceptional and that is the point. I joined Dr. Ethel Tshukudu's CS Education Research Lab at SJSU in August 2024 as an undergraduate researcher, starting on the ordinary work — study logistics, data handling, tooling. That became co-authorship on a peer-reviewed paper at the SIGCSE Technical Symposium 2026 — on bilingual coding for inclusive computer science learning (DOI 10.1145/3770761.3777339).
The sequence was: be useful on small things, be reliable, stay long enough for a project to reach a submission deadline. There was no shortcut in it, and there did not need to be. The single biggest determinant was time in the lab, not talent — projects reach submission deadlines on their own schedule, and you have to still be there when one does.
More on what I work on is on the research page and the about page. If the research is in CS education specifically, the bilingual coding guide covers one of these studies in detail, and CODESWITCH is what came out of it in practice.
FAQ
Shortlist five to eight faculty from the department directory whose work you can follow, read the abstract and introduction of two of their recent papers, check their lab site for a stated application process, and ask instructors who already know your work. Look outside your own department too — CS skills are in demand in education, biology, linguistics, and public health labs where competition is much thinner.
Under 200 words, naming one of their papers exactly and one specific thing in it you found interesting, stating a concrete skill and a real project where you used it, explicitly offering to do the unglamorous work, and asking for fifteen minutes rather than a position. Attach a one-page CV, follow up once after about ten days, and email several people because the reply rate is genuinely low.
Volunteer, course credit through an independent study, an hourly or funded assistantship, a structured summer programme with a stipend, and a supervised independent project. They differ in commitment, cost, and eligibility, and funded programmes in particular often carry conditions on citizenship, residency, or work authorisation that should be the first line you read.
In three passes. The first covers title, abstract, introduction, headings and conclusion in five to ten minutes and tells you what is claimed. The second covers figures, tables and the method in about an hour and tells you whether the evidence supports it. The third reconstructs the work and takes half a day, and is only worth it for papers you intend to build on.
Almost always something unglamorous: literature screening, data cleaning, transcription, coding responses against a codebook, running participant sessions, or building a small internal tool. Completing it well and on time is the entire mechanism by which undergraduates progress toward authorship, because reliability is a lab's scarcest resource.
An Institutional Review Board is the university committee that reviews research involving human participants before data collection begins. You need it for surveys, interviews, classroom studies, and usability sessions, and data collected without approval generally cannot be used or published. Between training, drafting, review and at least one revision round, six to twelve weeks before the first participant is a realistic plan, though timings vary widely by institution.
It depends on the field. In most of computer science the order reflects contribution, with the first author having done the most and the last position commonly held by the supervising principal investigator, while in theory, mathematics and economics alphabetical order is standard and carries no ranking at all. Ask what would earn authorship at the start of a project rather than at submission, and treat a named acknowledgement as a real and normal outcome.
Yes, most commonly as a poster, a workshop paper, a short paper, or a middle-author position on a full paper. It is rare enough among undergraduates to be memorable in any application pile, and the process teaches scoping a claim to the data, writing for hostile reviewers, and handling criticism — none of which coursework covers.