Read the technical posting and calendar together.
SIParCS is a route into scientific computing at NSF NCAR in Boulder. The 2027 project page expects applications and project descriptions around November 3, 2026, but the official pages disagree on the January closing date. A student should see that disagreement before relying on a calendar reminder.
| Item | Published detail |
|---|---|
| Opening | Projects and applications expected around November 3, 2026. |
| Closing conflict | Project page: January 19, 2027. Tentative calendar: January 20, 2027, 11:59 p.m. MST. Actual posting should resolve the conflict. |
| Technical-program dates | Tentatively May 17–July 30, 2027; 40 hours weekly, in person/hybrid in Boulder. |
| Academic stage | Enrolled at a U.S. university; continuing undergraduate/graduate conditions apply. May 2027 graduates entering graduate school may apply to graduate positions. |
| 2027 hourly rates | Published as $23.41 for undergraduates and $28.32 for graduate students. |
| Support | Housing, travel funding and regional transit pass are described; check the actual offer and arrangements. |
The technical program and the separate Outreach, Development and Education, or CODE, position also have different work and eligibility. Do not transfer CODE’s citizenship language or initial-consideration deadline to every technical role.
Pick a project by the task you want to learn.
A scientific-computing internship can involve performance, data preparation, software reliability or tools used by researchers. Start by naming the project’s input and output. Does it take a large simulation result and make analysis easier? Does it test a computation under a different configuration? Does it improve how somebody finds or uses data? These questions help you connect your experience to the actual work.
Choose evidence at the right scale. A small project where you measured runtime under controlled conditions can be relevant to performance work. An analysis with clear data definitions can be relevant to scientific data support. Explain the limitations instead of pretending that running a laptop example gave you experience operating a supercomputer.
When the 2027 descriptions appear, compare a few plausible projects rather than applying indiscriminately. Read the particular prerequisites, the mentor’s stated goal and how your current preparation connects. A project-specific explanation gives a reviewer more to work with than a general claim that all computing topics interest you.
A benchmark needs a question, not just a faster number.
For an original practice example, imagine you rewrite a function and its runtime falls from 2.0 seconds to 1.2 seconds on one input. That is a 40% reduction in observed elapsed time for that run. It does not establish a 40% improvement for every input or an improvement in scientific correctness.
Scientific-computing work-sample note
Test that both versions return the same intended result before comparing speed. Repeat the timing under a stated environment and try more than one input size. Record what changed, how timing was measured and whether setup or file loading is included. The useful evidence is the method and its limits, not the most favorable single measurement.
Write the result as a small report another person can reproduce. A graph may help, but the code, input definition and test instructions matter first. If the new version fails on a boundary case, document the failure rather than hiding that input from the comparison. This exercise is independent preparation, not an NCAR assessment.
Preserve hourly pay and the actual paid-hours rules.
The eligibility page labels its rates as 2027 and lists three observed holidays as unpaid for interns. That means a simple weekly-hours multiplication may overstate gross pay. Keep the hourly rate, approved work dates and actual paid hours separate rather than advertise a summer total the source does not promise.
For a hypothetical budget, a student offered 400 paid hours at $23.41 would calculate $9,364 before deductions. That is an arithmetic example, not a SIParCS total-hours commitment. Housing and travel support may materially affect affordability, but neither is equivalent to unrestricted wages. Check deposits, meal costs and when travel arrangements are made.
International students at U.S. universities are welcomed on the technical eligibility page, subject to appropriate work authorization. Use the program and your designated school official to resolve your specific authorization and timing. Do not infer eligibility from being able to submit a form, and do not transfer the more restrictive CODE wording to technical opportunities.
Resolve the deadline conflict with the posting, not a guess.
When the actual 2027 job appears, save its deadline and any specific instructions for documents or project preferences. Compare them with both program pages. If the inconsistency remains, ask the program which deadline controls. A personal target on or before January 19 can reduce the risk of waiting too long, but it should not be labeled a confirmed employer deadline.
The eligibility page also retains a reference to December 2025 graduates among otherwise 2027 material. Applicants whose graduation timing does not fit the clear current examples should ask for clarification. Do not mechanically replace every old year with 2027; that could create an eligibility rule the program never published.
Finish the application with an accurate project explanation, accessible evidence and the documents requested by the actual role. Then retain the submission confirmation. An attractive stipend and housing package are reasons to investigate the opportunity, not reasons to ignore a schedule or academic condition that makes it unsuitable.