Atlassian surveyed 3,500 developers and managers across six countries for its 2025 developer experience research and found that 90% lose six or more hours a week to organisational inefficiency, and 50% lose ten or more. The instructive part is what they blamed: finding information, adapting to new technology, context switching between tools, and collaborating with other teams. Coding, as Atlassian notes, “wasn’t listed in 2024 or 2025 as a source of time wasting.” If you are about to hand a remote intern a greenfield feature, you are scoping them into the one part of the job that was never the bottleneck.

Key takeaways

  • Writing code stopped being the constraint; finding context and reviewing changes became it.
  • Large pull requests stall — AI-assisted ones wait 16+ hours for pickup and merge at 32.7% within thirty days.
  • A good intern project is bounded, measurable before it starts, and off the critical path.
  • Scoping is also screening: how someone decomposes work is the signal you are actually buying.

The bottleneck moved, and most intern briefs did not

LinearB’s 2026 benchmarks are drawn from 8.1 million pull requests across 4,800 teams in 42 countries, and they describe a pipeline that has gone lopsided. Some 88.3% of developers now use AI regularly, up from just under 72% in early 2024. Code volume followed: AI-assisted pull requests run past 400 lines against 157 lines for unassisted work.

Review did not follow. Those same AI-assisted pull requests wait more than sixteen hours before a reviewer picks them up, against roughly 200 minutes for unassisted work, and only 32.7% of them merge within thirty days — against approximately 84.5% for human-authored changes. Generation scaled; the human attention that has to sign off did not.

For a twelve-person startup, that is the whole argument about intern scoping. Every hour of intern output has to pass through a senior engineer, and that engineer is the scarce resource. A brief that maximises lines written and minimises reviewability is a net withdrawal from your team, however talented the intern is.

Why the greenfield feature is the wrong first project

The instinct is generous: give the intern something real, something with their name on it. In practice a new feature is the worst-shaped work you can hand a newcomer. It requires the most context, which they have least of. It produces the largest diffs, which review slowest. It sits on the roadmap, so when it slips it slips visibly, and the intern’s six weeks become a story about a delay.

The friction layer inverts every one of those properties. It is discoverable without a product briefing, it decomposes naturally into small changes, and nothing downstream is waiting on it. It is also where your measured pain already is: the four things Atlassian’s respondents named cost real hours every week, and almost none of them require product judgment to fix.

intern-project-scoping-inline

What a well-scoped intern project looks like

Four properties, and a project that misses any one of them tends to end in a polite write-off at week six.

  • It has a number attached before it starts. Setup time for a new machine, open documentation gaps, coverage on one module, minutes of pipeline runtime. Measurable today, measurable again at the end.
  • It arrives in small pieces. A cadence of changes a reviewer can read in one sitting beats one large branch, for exactly the reason the pickup-time data shows.
  • Nothing on the roadmap is waiting for it. Off the critical path is not a demotion; it is what makes the work safe to give to someone you are still evaluating.
  • It is real. A sandbox exercise nobody merges teaches you nothing about the candidate and gives them nothing to point at.

A scoping checklist for the first six weeks

  1. Ask your team where the hours go. One question in Slack: what wasted your time this month? Take the top three answers as your candidate project list.
  2. Pick the one with a number. If nobody can state a before-value, choose a different item rather than inventing a metric later.
  3. Cut it into six weekly slices. Week one must end in something merged, however small — that first merge is the fastest signal you will get about whether this hire works.
  4. Cap the change size. Say the number out loud in the brief. Small diffs are the practice that keeps review from becoming the bottleneck.
  5. Commit to a review turnaround. Name a window — same working day, next morning — and hold to it. Across time zones, an unreviewed branch costs a full day of momentum.
  6. Ask for the reasoning, not just the diff. Two or three sentences per change on what was tried and rejected. That is where judgment is visible, and where a fabricated contribution falls apart.
  7. Re-measure at week six. Compare the number to where it started, and write down what you would scope differently. That note is the asset for the next intern.

Scoping well is a hiring decision, not an admin task

Six weeks of bounded, reviewed, measured work tells you more about a candidate than any interview loop, and it leaves an artifact either way. It is also a funnel: NACE’s national survey found the average intern conversion rate reached 63.1%, its highest in five years, and Handshake’s index found 82% of interns who felt fairly compensated would likely accept a full-time offer, against 63% of those who did not. Pay properly, scope tightly, and the internship is a hiring pipeline rather than a favour. We have written the operational half of this separately, as a six-week work trial and a first-30-days onboarding plan.

What scoping cannot solve is who arrives. Tierones is an Tier-1-only network of second- and third-year developers from India’s IITs, NITs and IIITs, verified by institute email, showing work that was checked rather than claimed — commits, shipped projects, ratings — and published as a verified profile. They work remotely, part-time, alongside their semester, and you set your own rate in USD. It is free while the network grows.

FAQ

What should a remote intern’s first project be?

Something at the friction layer rather than a greenfield feature. Atlassian’s 2025 survey of 3,500 developers and managers found 90% lose six or more hours a week to organisational inefficiency — to finding information, context switching and cross-team collaboration, with coding named in neither 2024 nor 2025. That work is bounded, reviewable, and valuable in the week it lands.

Why not give an intern a real feature to own?

Because features consume the resource you are shortest of: review attention. LinearB’s benchmarks across 8.1 million pull requests found AI-assisted changes run past 400 lines against 157 for unassisted work, wait over sixteen hours for a reviewer, and merge within thirty days only 32.7% of the time against roughly 84.5% for human-authored code. Large speculative branches stall; small bounded ones ship.

How do I know if the project was worth it?

Define the before-and-after number before the intern starts — setup time, documentation gaps, coverage on one module, pipeline runtime. Measure it on day one and again at week six. If you cannot name that number in advance, the project is not scoped yet, and no amount of intern effort will compensate for that.

Hiring a remote intern this quarter? Tell us what you need — we’ll show you proof, not CVs.