- From
- The ASCEND Lab · Virginia Tech CS
- To
- Undergraduates who want to do research, not watch it
- Re
- Joining a project team
Undergraduate research at ASCEND
You join a project team as an apprentice researcher — real responsibilities, alongside PhD and master's students, on work that ships and gets published.
Three things we mean by that: it's apprenticeship, not observation — you learn the craft by doing it. Teams are vertically integrated — someone is always one step ahead of you. And it's built for the long game — the best runs span several semesters and end with signature work: a paper, a tool, a poster with your name on it.
Filed under · how projects work
Project charter — what every team agrees to
Every project runs on the same four clauses, so you always know who you're working with and what “done” means this semester.
Two mentors, one team
A faculty lead sets the big-picture direction; a graduate or senior-peer mentor works with you directly, week to week.
Semester-sized milestones, multi-year arc
Work is scoped so one term is achievable — and laddered so the arc rewards staying.
Sub-teams own real pieces
Bigger projects split into sub-teams that each own a concrete slice and sync with the whole group.
Expectations in writing, week one
You and your mentor agree on goals, hours, and deliverables at the start. No ambiguity about what's expected.
Pinned to the wall · studio day
Studio day
Open studio — drop in
The first studio of every semester is an open house. Come see the lab, meet the teams, ask anything. No commitment required.
day + time set each semester · watch this space
The rhythm of the lab is the weekly studio session — the standing hour the whole team is in the room. You demo what you did, get feedback, pair with your mentor or sub-team, and leave with a concrete next task. It's also where you absorb the field — reading groups, practice talks, other people's problems — not just your own slice of it.
| Week of — | What | hrs |
|---|---|---|
| Studio day | All hands — demos, unblocking, pairing | ~1 |
| Your own time | Independent work on your task | ~3 |
| As needed | Mentor check-in, reading group | ±1 |
Margin notes · advice before you sign on
Guidance
Start early, stay a while
Students who begin earlier and stick with a project across terms get the most out of it — and the best outcomes.
multi-semester > one-and-done
Own your goals
Come with what you want out of it — skills, a paper, grad-school prep — and set goals with your mentor.
mentoring is a two-way doc
Research runs slower than coursework
Longer turnarounds and dead ends are the normal texture of research, not a sign you're failing.
dead ends ≠ failing
Build research skills deliberately
Read the literature, question methods, present at symposia when you're ready. Owning the whole process is the point.
the process is the product
Prerequisites vary by project — each open project lists what's good to have. Not sure whether you're a fit? Email David at dhsmith4@vt.edu, or drop into the next open house.
Stapled · further reading
- Council on Undergraduate Research — Characteristics of Excellence
- Ten simple rules for an undergraduate-intensive research lab
- Computational Apprenticeship — a CS research-lab case study
- Vertically Integrated Projects (VIP) — Georgia Tech
- Effective mentoring practices — UCF Office of Undergraduate Research
In the folder · one register page per project — flip with the tabs, tear a page off to sign on
Open Projects
Loading…
Open projects are members-only
Everything above is open to everyone — the project list itself is for approved ASCEND members. Sign in with your Virginia Tech Google account; an admin approves new accounts before you can browse and sign on.
Your account is pending approval
An admin needs to approve your account before you can see project details. Check back soon, or reach out to dhsmith4@vt.edu.
Purplex Development for SPLICE Interoperability
RecruitingPurplex is the lab's own problem-authoring platform, and SPLICE is the shared exchange format the broader CS-ed tooling community is converging on. You'd work on making Purplex speak SPLICE — exporting problems and interaction logs in a format other researchers' tools can read — and testing that round-trip against real SPLICE-compatible tools.
Good to have: Comfortable writing JavaScript/TypeScript. No research experience required.
Your team
David H. Smith IV
Assistant Professor & PI
Materials & resources
Materials coming soon.
Qualitative Analysis of Student Interactions with Probeable Problems
RecruitingWorking with Vanshika, you'd help create new "probeable problems" in Purplex — problems designed to be explored through questions rather than solved outright — then run think-aloud interviews where students work through them out loud while you observe. You'd help build the codebook for those interactions and look for patterns in how students probe unfamiliar problems.
Good to have: No coding required — this is a qualitative, participant-facing role.
Your team
David H. Smith IV
Assistant Professor & PI
Vanshika Punekar
Undergraduate Researcher
Materials & resources
Materials coming soon.
Decomposition Diagram Analysis and Autograding
RecruitingWe've collected a large set of decomposition diagrams from a data-science course — visual breakdowns of how students structured their solutions. Working with Monisha, you'd sort and categorize the diagram types students produced, help build a rubric for grading them consistently, and then prototype ways to auto-grade diagrams against that rubric.
Good to have: Comfort with organizing data (e.g. spreadsheets); Python is a plus for the autograding half but not required to start.
Your team
David H. Smith IV
Assistant Professor & PI
S. Moonwara A. Monisha
PhD Student
Materials & resources
Materials coming soon.
Investigating Grading and Feedback Approaches for EiPL and Prompt Problems
RecruitingEiPL ("Explain in Plain Language") and Prompt Problems are two exercise types the lab uses to study how students reason about code and AI-generated solutions. You'd help compare grading and feedback approaches for both — building a hand-graded baseline, testing an LLM-based grader against it, and analyzing where the two agree and diverge.
Good to have: Python and basic data analysis (e.g. pandas) will help; we'll teach the rest.
Your team
David H. Smith IV
Assistant Professor & PI
Materials & resources
Materials coming soon.
Designing "Debuggers" for Code Comprehension
RecruitingWe're prototyping a visual debugger — in the spirit of Python Tutor — built for code comprehension, not just bug-hunting. You'd help build the prototype, run sessions where students use it to make sense of unfamiliar code, and turn what you observe into concrete design goals for the next iteration.
Good to have: Comfortable in Python; front-end/JS experience is a plus for the prototyping half.
Your team
David H. Smith IV
Assistant Professor & PI
Materials & resources
Materials coming soon.