Common App activities-section examples for research: lab assistants, independent researchers, summer-program alumni, science competition finalists. Each one names the lab, the artifact, and the validation signal.
Position / leadership
Summer Researcher, MIT Primes Program
Organization
MIT Department of Mathematics
Description
Co-authored paper on combinatorial game theory; presented at Joint Mathematics Meetings. Paper accepted in Discrete Mathematics.
Why it works
Three independent proof points: peer-reviewed paper, conference talk, journal acceptance. Tier defaults to international because the journal does.
Position / leadership
Research Assistant, Pediatric Oncology Lab
Organization
Children's Hospital of Philadelphia
Description
Analyzed 2,400 patient outcomes in R. Co-presented at Pediatric Oncology Symposium. Lead author on poster at AACR.
Why it works
Real research artifacts: 2,400-patient dataset, R analysis, AACR poster. Hospital + AACR signals authentic, vetted research environment.
Position / leadership
Independent Researcher (Regeneron STS Top 300)
Organization
Self-directed; mentored by university PI
Description
Built CNN for early-stage melanoma detection. 94.2% sensitivity on 12,000-image holdout. Regeneron STS Top 300 in 2026.
Why it works
Independent research is a hard sell without external validation. Regeneron Top 300 IS that validation. The metric (94.2%, n=12,000) makes the claim auditable.
Position / leadership
Summer Intern, NIH High School Summer Program
Organization
NIH Bethesda, Section on Comparative Behavioral Genomics
Description
Genotyped 600 zebrafish over 8 weeks. Co-authored protocol now used by 4 NIH labs. Got my PI's recommendation for second summer.
Why it works
Lab name is specific. Protocol re-use across 4 labs is the artifact that beats hours logged. "Got asked back" is the strongest research signal.
Position / leadership
Finalist, Google Science Fair
Organization
Google Science Fair, top 24 globally
Description
Designed low-cost ($12) water-arsenic test using paper microfluidics. Filed provisional patent. Featured in WIRED magazine.
Why it works
Cost ($12), method (paper microfluidics), and validation (patent, press) are independent. Each could carry the line on its own.
Position / leadership
Research Assistant, Stanford NLP Group
Organization
Stanford University, supervised by Prof. Manning
Description
Improved tokenizer training speed by 1.8x using parallel batching. PR merged to Hugging Face Transformers. Got PyTorch credit.
Why it works
Specific lab and PI. Outcome is measurable (1.8x speedup) AND public (merged PR). PyTorch credit is a verifiable detail.
Paste your 10 Common App entries and Kolly scores each one on impact, specificity, and voice. Free to start.
Try the activity reviewer free