sndsh404/summer-2027-internships: a US-only, one-maintainer list for the 2027 cycle

What the list actually covers, and who it leaves out

This repository is a Markdown table of internship openings for Summer 2027 and off-season terms (fall, winter, spring), restricted to US-based positions, with US-based remote counted as in scope. The scope section names five role families: software engineering, data and ML, hardware, quant, and product. The rows in the README match that: Susquehanna and Point72 for quant, Google and Uber and The Trade Desk for software engineering, Databricks and Capital One for product, Neuralink for BCI applications, Second Order Effects for electrical engineering, Western Digital for a winter co-op.

The audience is narrow on purpose. The about section says the maintainer is US-based and built the list for himself, that it is US only for now, and that Canada might be added if the project grows. There is no search box, no filter by term, no email digest. You read a table.

That narrowness is the feature. A student who wants a Summer 2027 software internship in the US can scan the Added column, which is sorted newest first, and see what moved this week. A student looking for a marketing or accounting internship will find nothing, because those are explicitly out of scope. Related searches like "summer 2027 internships marketing" and "summer 2027 internships accounting" land on a page that cannot help them.

How the table, the legend and the Added column work together

The whole mechanism is a single Markdown table with five columns: Company, Role, Location, Apply, Added. The Apply cell is a link to the employer's own job portal, which is where the actual application happens. Nothing is hosted here.

The legend is where the useful metadata lives. Four markers appear inline in role names: a closed padlock for closed postings, a passport control symbol for roles that do not offer sponsorship, a US flag for roles that require US citizenship, and an hourglass for postings that state no term or year, described in the README as a rolling or evergreen req where you should confirm the term before applying. The Unison (GE Aerospace) engineering intern row carries both the no-sponsorship and citizenship markers. The Neuralink row carries the hourglass.

The Added column is the part worth understanding before you trust it. The README states plainly that the maintainer tried to pull a posted date from each job portal, that many portals (Greenhouse and smaller career sites are named) do not show one publicly, and that only some rows are dated while the rest stay a dash. New roles get a date where the portal shows one, and the README invites readers to open a PR with a date for an undated row. So the sort order is real for dated rows, and silently incomplete for the rest. If you are triaging by recency, undated rows are a blind spot, not a signal that they are old.

Getting the list and starting with the tracker spreadsheet

There is nothing to install. The repository is a README plus a spreadsheet, so the workflow is clone or download, then open the file.

git clone https://github.com/sndsh404/summer-2027-internships.git
cd summer-2027-internships
ls

The top level of the repository contains CONTRIBUTING.md, LICENSE, README.md and internship_tracker.xlsx, so the directory listing should show those four entries. If you only want the table, opening README.md in any Markdown viewer is enough.

The tracker is the one piece of tooling here. The README says to download it, upload it to Google Sheets, and bookmark it, and that it also works in Excel. It exists because, in the maintainer's words, he added it so people can keep all their applications in one place instead of losing track of everything.

# after cloning, the spreadsheet is in the repository root
ls internship_tracker.xlsx

From there the loop is manual: pick a row from the README table, follow the Apply link to the employer's portal, apply, and log the entry in the tracker. The README does not document how the spreadsheet's columns are laid out, so open it before you build a habit around it.

The maintenance model is one person, and the README says so

The README states that the repository is maintained by just the author, asks readers to contribute, and gives two concrete asks: report dead links in an issue so they get removed, and open an issue or PR for missing roles so they get added. It also says the author has declined offers to buy the repository, on the reasoning that it would get worse under a company's website with labels and ads.

The most useful sentence for planning is the one about scale. The README says the author is doing his part until the big repositories start posting 2027 roles officially, and that at that point it gets almost unmaintainable for one person. That is an explicit statement about the project's expected lifespan, not a marketing line. Treat this list as a bridge for the early part of the cycle, not as the thing you will use through the spring.

The last push to the repository was on 2026-08-03. The README describes an intent to update daily, and that intent is not the same as a guarantee. There are no releases to follow, so the only way to know whether a given row is current is to open the employer's link and check whether the posting is still live.

Where this list breaks down for a real search

The failure modes are mostly geographic and categorical, and they are stated rather than hidden. US only means Toronto and Singapore searches are out of scope, and the README says Canada is only a possibility if the project grows. The five role families exclude marketing, accounting, government and mechanical engineering, which are all real search terms people use around this cycle.

There is a second failure mode in the data itself. Because the Added column is blank for many rows, you cannot sort the table by freshness, and you cannot tell whether a row with a dash was added yesterday or three weeks ago. The README also notes that the legend's hourglass marker means the posting states no term or year, so a row can be in the list and still not be confirmed as a Summer 2027 role. The README's own instruction is to confirm the term before applying.

Finally, the list is a pointer, not a source of truth. Every Apply cell goes to an external portal that can close, change requirements, or stop accepting applications without the README changing. Dead links are exactly the thing the maintainer asks readers to report, which tells you they happen.

How it differs from the large aggregator repositories

The obvious alternative is one of the big community internship repositories that cover many cycles and countries with a large contributor base. The difference is not the format, since both are Markdown tables, but the editorial scope.

A large aggregator optimizes for coverage: more countries, more role types, more cycles, and a contribution pipeline deep enough to absorb the volume. This repository optimizes for a single reader's decision. It is US only, it names five role families, and it carries inline markers for sponsorship, citizenship and unstated terms that a broader list may leave to the posting text. The README's own framing is that it was started for personal use and grew, and that it will step back once the large repositories post 2027 roles officially.

So the trade is coverage against curation density. If you are searching outside the US, or in a role family the scope section excludes, the aggregator is the right tool and this one is not. If you are a US student in one of the five families and you want a short list where the constraints are marked next to the role, this is faster to read than a larger table where you have to open every posting to learn the same thing.

Licence, contributions and what the MIT terms mean here

The repository is MIT licensed. For a list of links to third-party job postings, that is permissive in the ordinary way: you can copy the table, fork it, or republish it with attribution, and the licence text in LICENSE is the operative document. It says nothing about the postings themselves, which belong to the employers and live on their portals.

One thing worth separating: contributing to the list is governed by CONTRIBUTING.md, not by the licence. The README tells you to read CONTRIBUTING.md before adding a role, and gives the shape of a contribution in the same paragraph. Find a dead link, drop it in an issue. Find a missing role, open an issue or a PR. Have a posted date for an undated row, open a PR and it will be merged. Those are the three contribution paths the README names.

There is no CLA mentioned, no release process, and no versioned data file, so a fork is a full copy of the state at the last push on 2026-08-03. If you fork it and keep it current yourself, you inherit the maintenance work the README describes as hard for one person.

Editorial conclusion

Use this list if you are a US-based student targeting Summer 2027 or an off-season term in software, data, hardware, quant or product, and you want a dated view of what has opened rather than a search interface. Skip it if you need Canada, marketing, accounting, government or mechanical engineering roles, since the scope section excludes them, or if you need guaranteed daily updates. Before relying on it, check the Added column for the roles you care about, confirm the term on the employer's own posting, and read CONTRIBUTING.md so any issue you file or PR you open matches the format the maintainer asks for.

Frequently asked questions

Official sources

1. Issues
2. License: MIT
3. README
4. sndsh404/summer-2027-internships on GitHub