As a sophomore or junior, choose between research and building a product by your next step. Aiming for a PhD, make research the main line, because faculty letters and lab experience matter most there. Aiming for a job, a product shows applied skill. Give one about two-thirds of your spare hours and keep the other small, since both reward depth more than splitting time evenly.
Should a sophomore or junior do research or build a product?
You have maybe ten to fifteen spare hours a week after classes. One advisor says get into a lab. A friend who just landed an internship says build something. Both are right for someone. The question is which one is right for your next step, and how to avoid doing both badly.
There is no single page that answers this. What exists is scattered advice from professors, career offices and NSF. Read together, it points to a fairly clear rule.
What do professors and career offices agree on?
Depth beats spreading thin. Faculty who take undergraduates are explicit that supervision costs them time. Michael Gleicher, a computer science professor at the University of Wisconsin–Madison, writes on his advice page for undergraduates that mentoring a student takes substantial time from him and often from a graduate student too, so he takes on students only when a project genuinely needs help. Pieter Abbeel, a UC Berkeley electrical engineering and computer sciences professor, sets the bar in hours on his getting-involved page: full-time in the summer, and ideally 15 or more hours a week during the semester. Neither describes a lab where a few hours a week leads anywhere.
Projects count, especially early. Gleicher suggests most undergraduates aim for project experience first and treat research as a bonus. He encourages self-directed projects, such as games, interactive pieces or websites, posted on a web page, and he finds them more compelling than a list of finished courses. A career post from Ohio State's engineering career services, written by a peer career coach, makes the same case for the first couple of years: projects are among the strongest resume items when you have little professional experience, as long as you say exactly what you did. Its one-line version: "Projects stand out because they demonstrate applied knowledge."
Summer is where research happens. Abbeel only takes undergraduates who can work full-time that summer. NSF's REU page for students describes summer sites where a group of about ten students does real research with faculty, with application deadlines mostly between January and March, and materials that usually include reference letters. So the academic year is when you prepare and the summer is when research takes off.
Where do they disagree, or go quiet?
They disagree on order. Gleicher says projects first; his reasons are that you need the coursework before you can help, and that his own time is limited. Abbeel, when a student is torn for the summer, generally recommends an industry internship over research, with side projects alongside if research interest grows later. A student reading only one of them would get a different answer.
They are quiet on two things. None says how a product built on your own compares with research for a PhD application. And none addresses students who want both. That gap is what the rule below tries to fill. It is our reading of the sources, not something they state.
A decision rule: pick by your next step
| Your next step | Main line (about two-thirds of spare hours) | Side line (the rest) |
|---|---|---|
| PhD or research-focused master's | Research with one lab; aim for a summer REU or lab position | A small tool that supports the research, or none |
| Industry job or PM role | Your own product with real users | One class project or a short research rotation for breadth |
| Not sure yet | Sophomore year: a product, because you can start it today | Junior year: apply for a summer REU and test whether research fits |
Why two-thirds? Because the sources reward depth, and the main line needs enough hours to produce something a professor or recruiter can actually check. The side line keeps the other door open without stealing from the main one.
How to split the time, step by step
- Write down your next step in one sentence. "Apply to PhD programs in computational biology" or "get a product role after graduation." If you can't, use the "not sure" row.
- Count your real spare hours. Take a normal week, subtract classes, study time, work and sleep. Be honest; most students overestimate.
- Protect grades first. Labs and grad programs both read transcripts, and Abbeel names strong grades in difficult classes among the things that help an application.
- Give the main line fixed blocks. Two or three long blocks a week beat many short ones, for lab work and for building alike.
- Keep the side line deliberately small. One evening a week, with one concrete output per month.
- Set a checkpoint before winter break. If the main line produced nothing you can show by then, change something before REU deadlines arrive.
- Write it down every week. A short log of what you did, what broke and what's next becomes your interview material and the basis for any letter.
What does a product prove that research doesn't, and the reverse?
A product with real users proves that you can finish, ship and handle feedback from people who owe you nothing. Research proves that you can work on an open question, under someone who can later describe your work in a letter. Neither substitutes for the other in the setting where the other matters. If you have both, say which one was your main line and why. That honesty reads better than claiming equal depth in each.
How STARC fits this choice
STARC has two tracks that map onto the two rows. The one-person company cohort runs 8 weeks at about 10 hours a week: you build a small product with AI, find real users and write one build log a week. That fits as the main line for the industry path, or as one focused term for a student who isn't sure. The research track is small and opens only when a mentor genuinely has room, and publication is never promised. The rules for both are on the How it works page, and common questions are in the FAQ.
The first cohort hasn't run yet, so there are no results to point to. If you are torn between the two, use the decision rule above first; a program should fit your plan, not replace it.
The short version
Decide by your next step. Research for a PhD, a product for a job, and if unsure, build now and test research with a summer REU. Give the main line about two-thirds of your spare hours, keep grades safe, and log your work every week.
Related questions
Can I start research as a sophomore?
Often yes, but some professors want you to finish the relevant coursework first. A University of Wisconsin–Madison professor, for example, says students shouldn't start research with him until they have taken a relevant class. Ask the specific lab.
How many hours does undergraduate research take?
It varies by lab. One UC Berkeley professor asks undergrads to be available full-time in the summer and ideally 15 or more hours a week during the semester. Ask any lab you contact for its expectation before you commit.
When do I apply for summer research (REUs)?
According to NSF, most REU sites run in the summer and set application deadlines between January and March. You apply to each site directly, usually with a resume, statement, transcript and reference letters.
Does a product I built myself help a grad school application?
It can show initiative and that you finish things, and it gives you real material for interviews. For research programs, though, it does not replace research experience and letters from people who watched you do research.
Build something real
OPC cohort or research track: real work, judged by what you actually made.
Follow STARC on X. New posts, what we're building, and the questions students send us: @Starc_Institute
Follow on XLast updated: 2026-10-27. Written by the STARC team. Rules quoted here come from our how it works page; if they change, this post is updated.
More questions
How do I find my first 10 users on campus who aren't my friends?
How a college student can find 10 real users on campus beyond friends: pick a narrow group, go where it gathers, and ask the right way.
Monkey King and Lin Daiyu as a couple: what's the pairing, and is there a game?
Why fans paired the Monkey King with Lin Daiyu from Dream of the Red Chamber, and a free browser game where you plan the couple's days.
Can one person with no coding experience ship a web game with AI in a week?
What it took to build and launch a small web game with AI in five days: who did what, where image generation broke, and how to run with no server.