How Project Deep Dive Interviews Are Scored: Rubric by Level (2026)

A project deep dive interview is scored on 6 dimensions: context and own role (15%), technical depth (25%), decisions and trade-offs (20%), impact (15%), scope and influence for the level (10%), communication (15%). Each is rated 1 to 4 against anchors that rise with the level. At mid level, an outstanding score on technical depth means: "Goes three levels deep and explains internals of the tools they relied on."

The dimensions and their weights

DimensionWeightSolid (3) at mid level
Technical depth25%Survives "why" probes two or three levels deep on their own design.
Decisions and trade-offs20%Names alternatives considered, why they chose and what they would change.
Context and own role15%Explains the problem, users and constraints, and the part they owned end to end.
Impact15%States outcomes with numbers and links them to the user or business goal.
Communication15%Structured and concise, adjusting depth to the listener.
Scope and influence for the level10%Describes owning a feature or project end to end.

What changes between levels

On technical depth, the heaviest dimension, an outstanding mid-level answer "goes three levels deep and explains internals of the tools they relied on", while an outstanding senior answer "explains system internals, failure modes and limits in depth under sustained probing". The same answer can score 4 at mid level and 2 at senior: senior candidates are expected to raise things unprompted that mid-level candidates may be asked about.

Every score, at every level

Context and own role (15%)

Level1 · Poor2 · Borderline3 · Solid4 · Outstanding
Intern / new gradCannot explain what the college or internship project did or who it was for.Explains the project but cannot separate their part from the team's.Explains the project's purpose, its users and the parts they personally built.Frames the problem crisply and states exactly which modules and decisions were theirs.
JuniorCannot explain what the project did or who it was for.Explains the project but describes their role only as "worked on it".Explains the problem, the users and the tasks they personally delivered.Frames the problem and their role clearly, including what they owned end to end.
Mid-levelCannot explain what the project did or who it was for.Explains the project but needs probing to separate their work from the team's.Explains the problem, users and constraints, and the part they owned end to end.Frames the business problem and constraints first, then their ownership, in under two minutes.
SeniorCannot explain what the project did or who it was for.Describes their own components well but not the wider project or why it mattered.Explains the business context, the teams involved and the direction they personally set.Frames why the project mattered to the business and exactly where they shaped its course.
StaffCannot explain what the project did or who it was for.Describes the project well but their role reads as one team's lead, not wider.Explains the org context, the teams involved and the direction they set across them.Frames the strategic problem, why it was theirs to solve and how they shaped several teams' work.

Technical depth (25%)

Level1 · Poor2 · Borderline3 · Solid4 · Outstanding
Intern / new gradCannot explain how their own code or project works when asked.Explains the stack used but not how the main parts work.Explains how their part works and answers one level of "why" follow-ups.Answers two levels of "why" on their own design, such as why that database or library.
JuniorCannot explain how their own work functions when asked.Describes what was built but falters at the first "why" probe.Explains their components and survives two levels of "why" on them.Survives three levels of probing on their own components without bluffing.
Mid-levelCannot explain how their own work functions when asked.Answers surface questions but bluffs or stalls two levels down.Survives "why" probes two or three levels deep on their own design.Goes three levels deep and explains internals of the tools they relied on.
SeniorCannot explain how their own work functions when asked.Deep on their own components but shallow on how the system fits together.Survives deep probes on both their components and the system's key interactions.Explains system internals, failure modes and limits in depth under sustained probing.
StaffCannot explain how their own work functions when asked.Deep on details but cannot explain the architecture's limits or evolution.Explains the architecture, its limits and how it would evolve, surviving deep probes.Moves easily between architecture and code-level detail and names the next scaling limit.

Decisions and trade-offs (20%)

Level1 · Poor2 · Borderline3 · Solid4 · Outstanding
Intern / new gradCannot say why any choice was made beyond "the tutorial used it".Gives a reason for a choice only when asked, with no alternative named.Names one alternative they considered and why they chose differently.Names alternatives, why they chose and what they would change with hindsight.
JuniorCannot say why any choice was made.Explains choices only as given by someone else.Explains the reasons for their own choices and one alternative considered.Explains choices, alternatives and what they would do differently now.
Mid-levelCannot say why any choice was made.Names choices but cannot explain the alternatives or their costs.Names alternatives considered, why they chose and what they would change.Explains trade-offs with evidence, such as a benchmark or incident, and what they would change.
SeniorCannot say why any choice was made.Explains component-level choices but not the major design decisions.Explains major design trade-offs and why they fit this team's context and constraints.Explains major trade-offs, a decision they got wrong and how they corrected course.
StaffCannot say why any choice was made.Explains technical trade-offs but not their cost or organisational effects.Explains trade-offs in terms of cost, team capacity and long-term ownership.Explains trade-offs for this team, this quarter and this customer base, with hindsight.

Impact (15%)

Level1 · Poor2 · Borderline3 · Solid4 · Outstanding
Intern / new gradStates no outcome for the project at all.States a vague outcome such as "it worked" or "we got good marks".States a concrete outcome, such as users, a demo, a prize or a merged contribution.States a measured outcome, such as response time halved or a hundred active users.
JuniorStates no outcome for the project at all.States a vague outcome with no number or concrete result.States the result of their work with a number where one exists.States a measured result and how it was measured.
Mid-levelStates no outcome for the project at all.States outcomes vaguely, without numbers even when asked.States outcomes with numbers and links them to the user or business goal.States quantified outcomes, how they were measured and what decisions they drove.
SeniorStates no outcome for the project at all.Quantifies their component's results but not the project's business effect.Quantifies the project's business impact and their contribution to it.Quantifies business impact and explains the lasting change it caused beyond the project.
StaffStates no outcome for the project at all.Quantifies project results but not the wider organisational effect.Quantifies org-level impact, such as cost, revenue or reliability across teams.Quantifies org-level impact and the durable change in how teams build or operate.

Scope and influence for the level (10%)

Level1 · Poor2 · Borderline3 · Solid4 · Outstanding
Intern / new gradDescribes only following tutorial steps, with no independent contribution.Describes small assigned tasks within someone else's project.Describes building a meaningful part of a course, personal or internship project.Describes leading a project, or owning a substantial internship deliverable end to end.
JuniorCannot point to any part of the project they built themselves.Describes small tasks with close supervision throughout.Describes owning tasks within a project with light supervision.Describes owning a small feature end to end with little supervision.
Mid-levelDescribes only work others directed, with no part they owned end to end.Describes task-level work only, directed by others.Describes owning a feature or project end to end.Describes owning a project and influencing an adjacent team's work.
SeniorDescribes only individual tasks and no project-level ownership or decisions.Describes owning a single feature with no cross-team influence.Describes leading a multi-team project or mentoring others through it.Describes setting direction for a multi-team effort and growing other engineers.
StaffDescribes only own-team tasks, with no influence on direction beyond them.Describes leading one team's project with no wider influence.Describes setting technical direction across several teams.Describes org-wide influence, such as a platform or standard several teams adopted.

Communication (15%)

Level1 · Poor2 · Borderline3 · Solid4 · Outstanding
Intern / new gradCannot give a coherent overview of the project, even when guided.Explains in a jumbled order and needs steering to the key points.Gives a clear overview first, then answers questions directly.Gives a clear overview, uses a diagram or example and adapts to follow-ups.
JuniorCannot give a coherent overview of the project, even when guided.Dives into detail before giving any overview.Gives a structured overview and answers follow-ups directly.Gives a concise overview and adjusts depth to the interviewer's questions.
Mid-levelCannot give a coherent overview of the project, even when guided.Explains clearly but spends too long on background or minor detail.Structured and concise, adjusting depth to the listener.Leads with the headline, uses diagrams well and anticipates follow-up questions.
SeniorCannot give a coherent overview of the project, even when guided.Clear on detail but cannot summarise the project's essence briefly.Summarises the project crisply and goes deep only where the interviewer probes.Summarises crisply, anticipates probes and handles challenge with evidence.
StaffCannot give a coherent overview of the project, even when guided.Explains well to engineers but not in terms of business stakes.Explains for both technical and business listeners and stays on point.Tailors the story to the listener instantly and handles challenge with ease.

Get scored against this rubric

Thita's AI interviewer runs a live project deep dive round and scores you against exactly this rubric, dimension by dimension, for the level you choose.

Practise a project deep dive

Where this rubric comes from

These rubrics are researched: they follow the scoring methods big-tech companies describe publicly (behaviourally anchored ratings, bar raisers, leadership principles, named interview signals) and how their rounds are commonly run. No company publishes its actual scoring sheet, so the level-by-level wording is our own synthesis. It is the same rubric our AI interviewer scores every mock against.

Questions, answered

How is a project deep dive interview scored?
A project deep dive interview is scored on 6 dimensions: context and own role (15%), technical depth (25%), decisions and trade-offs (20%), impact (15%), scope and influence for the level (10%), communication (15%). Each is rated 1 to 4 against anchors that rise with the level. At mid level, an outstanding score on technical depth means: "Goes three levels deep and explains internals of the tools they relied on."
What is the difference between a mid-level and a senior project deep dive answer?
On technical depth, the heaviest dimension, an outstanding mid-level answer "goes three levels deep and explains internals of the tools they relied on", while an outstanding senior answer "explains system internals, failure modes and limits in depth under sustained probing". The same answer can score 4 at mid level and 2 at senior: senior candidates are expected to raise things unprompted that mid-level candidates may be asked about.
What does a score of 3 mean?
Scores are relative to the level you are interviewing for. A 3 means solid for that level: the bar is met without help on the key points. A 4 means the candidate also raised the important points unprompted. A 2 usually means the interviewer had to lead the candidate to them.
Is this the official rubric used by Google, Meta or Amazon?
These rubrics are researched: they follow the scoring methods big-tech companies describe publicly (behaviourally anchored ratings, bar raisers, leadership principles, named interview signals) and how their rounds are commonly run. No company publishes its actual scoring sheet, so the level-by-level wording is our own synthesis. It is the same rubric our AI interviewer scores every mock against.