Imagry Software Engineer Interview Guide

The 60-second brief

A Software Engineer at Imagry is at the absolute forefront of the autonomous vehicle revolution. Unlike traditional self-driving systems that rely heavily on high-definition (HD) mapping, Imagry pioneered a mapless autonomous driving technology. This unique approach means our software must perceive, understand, and navigate complex environments in real time, mimicking human cognitive processes. As a engineer here, your code directly impacts how safely and efficiently a vehicle makes split-second decisions on public roads. Whether you are optimizing real-time C++ code on edge hardware, training advanced deep learning models for Perception, or designing robust algorithms for Motion Planning, your contributions shape the core platform. The work is highly complex, interdisciplinary, and safety-critical.

This guide is scoped to a Software Engineer candidate at Imagry.

Focus areas

• C++ (Programming)
• Deep Learning
• Motion Planning

The Imagry interview process

Imagry candidates report 4 rounds over 3-5 weeks. The stages below are what candidates describe, not a published process.

Introductory Call

Initial call with a recruiter to discuss the candidate's background and the role.

What to demonstrate

• Initial call with a recruiter to discuss the candidate's background and the role
• Depth in C++ (Programming)

How to prepare

• Answer aloud and timed: Explain the difference between std::unique_ptr and std::shared_ptr and describe how they impact memory overhead in a multithreaded application.
• Answer aloud and timed: How does the virtual table (vtable) work in C++, and what is its performance cost during runtime?

Technical Assessments

A series of technical evaluations starting from initial screens to deep-dive sessions.

What to demonstrate

• A series of technical evaluations starting from initial screens to deep-dive sessions
• Depth in C++ (Programming)

How to prepare

• Answer aloud and timed: Describe how you would implement a lock-free queue for high-throughput, low-latency data transfer between a camera sensor thread and a processing thread.
• Answer aloud and timed: What is Resource Acquisition Is Initialization (RAII), and how does it prevent resource leaks in complex exception-handling scenarios?

In-Person Visits

Candidates may have multiple in-person visits to the office, especially for those near key engineering hubs.

What to demonstrate

• Candidates may have multiple in-person visits to the office, especially for those near key engineering hubs
• Depth in C++ (Programming)

How to prepare

• Answer aloud and timed: Explain the concept of cache locality and how you would design a data structure to maximize CPU cache hits.
• Answer aloud and timed: How do you address the challenge of object occlusion in real-time 3D object detection?

Leadership and HR Alignment

Final discussions to ensure alignment with leadership and HR before making an offer.

What to demonstrate

• Final discussions to ensure alignment with leadership and HR before making an offer
• Depth in C++ (Programming)

How to prepare

• Answer aloud and timed: Describe the architectural differences between single-stage and two-stage object detectors, and how you choose between them for edge deployment.
• Answer aloud and timed: How would you handle class imbalance in a dataset where rare road obstacles (like construction cones or animals) appear infrequently?

Where candidates lose points

Demonstrate Seniority through Ownership

If you are interviewing for a senior role, don't wait for the interviewer to prompt you about edge cases, testing, or system constraints. Proactively discuss how you would structure your code for production, write unit tests, and monitor the performance of your system in the field.

Master Your Tools

Be ready to write syntactically correct code without relying on IDE auto-complete. If you are a C++ candidate, know your standard library (STL) inside and out. If you are a deep learning candidate, be prepared to write custom training loops or loss functions on a whiteboard.

Understand the Hardware Constraints

Autonomous driving software does not run in an infinite-resource cloud environment. Always frame your technical answers with hardware constraints in mind. Mentioning CPU/GPU memory limits, latency budgets, and thermal throttling shows that you understand the realities of deploying robotics software in the physical world.

Questions & practice

Explain the difference between std::unique_ptr and std::shared_ptr and describe how they impact memory ove

Medium · C++ & Systems Programming

Explain the difference between std::unique_ptr and std::shared_ptr and describe how they impact memory overhead in a multithreaded application.

Approach

1. Say what the runtime actually does before reasoning about the code.
2. Name what is shared across threads and what owns each piece of state.
3. Identify the window where an invariant is briefly untrue.
4. Distinguish a value from a reference to it, and say which one you handed out.

Follow-up

• What happens if two callers reach this at the same time?
• Where could this allocate more than you expect?

How does the virtual table (vtable) work in C++, and what is its performance cost during runtime?

Medium · C++ & Systems Programming

Approach

1. Say what the runtime actually does before reasoning about the code.
2. Name what is shared across threads and what owns each piece of state.
3. Identify the window where an invariant is briefly untrue.
4. Distinguish a value from a reference to it, and say which one you handed out.

Follow-up

• What happens if two callers reach this at the same time?
• Where could this allocate more than you expect?

Describe how you would implement a lock-free queue for high-throughput, low-latency data transfer between a ca

Medium · C++ & Systems Programming

Describe how you would implement a lock-free queue for high-throughput, low-latency data transfer between a camera sensor thread and a processing thread.

Approach

1. Say what the runtime actually does before reasoning about the code.
2. Name what is shared across threads and what owns each piece of state.
3. Identify the window where an invariant is briefly untrue.
4. Distinguish a value from a reference to it, and say which one you handed out.

Follow-up

• What happens if two callers reach this at the same time?
• Where could this allocate more than you expect?

What is Resource Acquisition Is Initialization (RAII), and how does it prevent resource leaks in complex excep

Medium · C++ & Systems Programming

What is Resource Acquisition Is Initialization (RAII), and how does it prevent resource leaks in complex exception-handling scenarios?

Approach

1. Say what the runtime actually does before reasoning about the code.
2. Name what is shared across threads and what owns each piece of state.
3. Identify the window where an invariant is briefly untrue.
4. Distinguish a value from a reference to it, and say which one you handed out.

Follow-up

• What happens if two callers reach this at the same time?
• Where could this allocate more than you expect?

Explain the difference between the A* algorithm and Dijkstra's algorithm. In what scenarios would you use a st

Medium · Motion Planning & Robotics Algorithms

Explain the difference between the A* algorithm and Dijkstra's algorithm. In what scenarios would you use a state-lattice planner instead?

Approach

1. Restate the input: its shape, its size, and what is guaranteed about it.
2. Name the brute-force solution and its complexity before improving on it.
3. Choose the data structure from the access pattern, not from familiarity.
4. State the target complexity and say which constraint rules the naive version out.

Follow-up

• How does this change if the input no longer fits in memory?
• What is the worst case, and how likely is it on real data?

How do you formulate a trajectory optimization problem using Model Predictive Control (MPC) under dynamic vehi

Medium · Motion Planning & Robotics Algorithms

How do you formulate a trajectory optimization problem using Model Predictive Control (MPC) under dynamic vehicle constraints?

Approach

1. Restate the input: its shape, its size, and what is guaranteed about it.
2. Name the brute-force solution and its complexity before improving on it.
3. Choose the data structure from the access pattern, not from familiarity.
4. State the target complexity and say which constraint rules the naive version out.

Follow-up

• How does this change if the input no longer fits in memory?
• What is the worst case, and how likely is it on real data?

How do you model and account for the uncertainty of dynamic obstacles (such as pedestrians or other vehicles)

Medium · Motion Planning & Robotics Algorithms

How do you model and account for the uncertainty of dynamic obstacles (such as pedestrians or other vehicles) in your path planning?

Approach

1. Restate the input: its shape, its size, and what is guaranteed about it.
2. Name the brute-force solution and its complexity before improving on it.
3. Choose the data structure from the access pattern, not from familiarity.
4. State the target complexity and say which constraint rules the naive version out.

Follow-up

• How does this change if the input no longer fits in memory?
• What is the worst case, and how likely is it on real data?

Explain how kinematic and dynamic vehicle models differ, and why dynamic models are necessary at higher drivin

Medium · Motion Planning & Robotics Algorithms

Explain how kinematic and dynamic vehicle models differ, and why dynamic models are necessary at higher driving speeds.

Approach

1. Restate the input: its shape, its size, and what is guaranteed about it.
2. Name the brute-force solution and its complexity before improving on it.
3. Choose the data structure from the access pattern, not from familiarity.
4. State the target complexity and say which constraint rules the naive version out.

Follow-up

• How does this change if the input no longer fits in memory?
• What is the worst case, and how likely is it on real data?

Write the update path that detects a concurrent edit

Medium · optimistic concurrency · transactional outbox · audit history

resource carries version INT NOT NULL DEFAULT 1. resource_revision holds revision_id, resource_id, version, actor_user_id, change_kind, patch JSONB, request_id, created_at with UNIQUE (resource_id, version). outbox_event holds aggregate_type, aggregate_id, aggregate_version, event_type, payload, status. A PUT carries the version the client read. Write the exact statements for the single transaction that applies the edit, records the revision and enqueues 'resource.updated', and give the handler's branch on zero affected rows. Then say what PostgreSQL 16 does under READ COMMITTED when two of these updates hit one row at once.

Approach

1. One transaction, three writes, no network call inside it: UPDATE resource SET title = $3, version = version + 1, updated_at = now() WHERE resource_id = $1 AND tenant_id = $4 AND version = $2; then INSERT the resource_revision row at version $2 + 1; then INSERT the outbox_event row at the same aggregate_version. The event goes to a table rather than a broker because no transaction spans both.
2. Branch on the affected-row count before doing anything else. Zero has three causes — stale version, wrong tenant, row gone — so re-read once and map to 409 carrying the current version, or 404 for an id outside the caller's tenant, which also stops the endpoint confirming that another tenant's id exists.
3. State the engine behaviour instead of assuming it. Under READ COMMITTED the second UPDATE blocks on the row lock, and when the first commits PostgreSQL re-evaluates the WHERE clause against the newly committed row, so the version predicate now fails and the statement reports zero rows. Under REPEATABLE READ the identical collision raises SQLSTATE 40001 instead, so the handler must fold both shapes into one conflict response.
4. Keep UNIQUE (resource_id, version) even though the predicate already serialises writers. It is what makes a lost update unwritable if any other path ever reaches the revision table, and it converts a logic bug into 23505 rather than into a silently missing history row.

Follow-up

• A client sends the version it read ten minutes ago and the resource has moved three versions. What is in your 409 so it can resolve the conflict without a full re-fetch?
• Two editors, two disjoint fields, no overlap. Does your answer still refuse the second write, and should it?

Explain why the owner filter ignores the listing index

Medium · composite index · query plans · sort elimination

The only index on resource is (tenant_id, status, updated_at DESC, resource_id DESC). A new endpoint returns one user's resources across all statuses, newest created first: WHERE tenant_id = $1 AND owner_user_id = $2 ORDER BY created_at DESC LIMIT 20. On a tenant with 2M rows it takes 900 ms and EXPLAIN shows a sort above a large scan. Explain precisely why the existing index cannot serve it, give the index that can, and state which of these the new index still will not help: owner_user_id alone across tenants; the same query ordered by updated_at. PostgreSQL 16.

Approach

1. Separate the two jobs an index does. For filtering, a composite btree is seekable only on a left prefix, so with no predicate on status the scan can at best range over tenant_id and test owner_user_id per row; PostgreSQL 16 has no btree skip scan to jump the unconstrained column.
2. For ordering, the index is sorted by (status, updated_at) within a tenant and not by created_at, so the LIMIT cannot stop early: every matching row is read and then sorted. That is the 'Sort Method: top-N heapsort' line, and it is why the plan reads 2M rows to answer with 20.
3. Derive the replacement from the access path — equality, equality, then the ordering column: CREATE INDEX CONCURRENTLY ON resource (tenant_id, owner_user_id, created_at DESC). The scan seeks to the (tenant, owner) range and walks 20 entries in order, so the Sort node disappears along with the row-read.
4. Treat INCLUDE (title, status) as conditional, not free. An index-only scan still visits the heap for any row whose page is not marked all-visible, so on a table taking 1.2k writes/second the win depends on autovacuum keeping the visibility map current, and the wider index costs more on every insert.

Follow-up

• 90% of rows are status='active'. Would a partial index WHERE status = 'active' change your answer, and for which of the three queries?
• A dashboard runs this for 40 owners in one page load. What changes about the design?

Explain the concept of cache locality and how you would design a data structure to maximize CPU cache hits.

Medium · C++ & Systems Programming

Approach

1. Fix the scope first: who calls this, how often, and what they do when it fails.
2. Name the read and write paths separately; they rarely have the same bottleneck.
3. Choose a partition key and say what query it makes expensive.
4. State the consistency you need, and where you are willing to be stale.

Follow-up

• What breaks first when traffic grows ten times?
• How does this behave when that dependency is down for an hour?

How do you address the challenge of object occlusion in real-time 3D object detection?

Medium · Deep Learning & Computer Vision

Approach

1. Clarify what is being asked and what a complete answer contains.
2. State your assumptions explicitly before working the problem.
3. Say what you would check first and why it is the highest-information step.
4. Work from the requirement backwards to the design.

Follow-up

• What assumption would you test first?
• How would you know your answer was wrong?

Describe the architectural differences between single-stage and two-stage object detectors, and how you choose

Medium · Deep Learning & Computer Vision

Describe the architectural differences between single-stage and two-stage object detectors, and how you choose between them for edge deployment.

Approach

1. Fix the scope first: who calls this, how often, and what they do when it fails.
2. Name the read and write paths separately; they rarely have the same bottleneck.
3. Choose a partition key and say what query it makes expensive.
4. State the consistency you need, and where you are willing to be stale.

Follow-up

• What breaks first when traffic grows ten times?
• How does this behave when that dependency is down for an hour?

How would you handle class imbalance in a dataset where rare road obstacles (like construction cones or animal

Medium · Deep Learning & Computer Vision

How would you handle class imbalance in a dataset where rare road obstacles (like construction cones or animals) appear infrequently?

Approach

1. Clarify what is being asked and what a complete answer contains.
2. State your assumptions explicitly before working the problem.
3. Say what you would check first and why it is the highest-information step.
4. Work from the requirement backwards to the design.

Follow-up

• What assumption would you test first?
• How would you know your answer was wrong?

Explain the mathematical principles behind camera calibration and how projection matrices are used to map 2D i

Medium · Deep Learning & Computer Vision

Explain the mathematical principles behind camera calibration and how projection matrices are used to map 2D image coordinates to 3D world space.

Approach

1. Clarify what is being asked and what a complete answer contains.
2. State your assumptions explicitly before working the problem.
3. Say what you would check first and why it is the highest-information step.
4. Work from the requirement backwards to the design.

Follow-up

• What assumption would you test first?
• How would you know your answer was wrong?

What strategies do you use to optimize a deep learning model for inference on low-power edge hardware without

Medium · Deep Learning & Computer Vision

What strategies do you use to optimize a deep learning model for inference on low-power edge hardware without sacrificing critical accuracy?

Approach

1. Clarify what is being asked and what a complete answer contains.
2. State your assumptions explicitly before working the problem.
3. Say what you would check first and why it is the highest-information step.
4. Work from the requirement backwards to the design.

Follow-up

• What assumption would you test first?
• How would you know your answer was wrong?

Describe how you would design a behavioral decision-making state machine to handle navigating a busy, unstruct

Medium · Motion Planning & Robotics Algorithms

Describe how you would design a behavioral decision-making state machine to handle navigating a busy, unstructured roundabout.

Approach

1. Fix the scope first: who calls this, how often, and what they do when it fails.
2. Name the read and write paths separately; they rarely have the same bottleneck.
3. Choose a partition key and say what query it makes expensive.
4. State the consistency you need, and where you are willing to be stale.

Follow-up

• What breaks first when traffic grows ten times?
• How does this behave when that dependency is down for an hour?

Describe a time when you had to debug a highly complex, intermittent technical issue in production. What was y

Medium · Behavioral & Professional Alignment

Describe a time when you had to debug a highly complex, intermittent technical issue in production. What was your systematic approach?

Approach

1. Establish what changed and when, before forming any theory.
2. Pick a bisection that eliminates candidates whichever way it turns out.
3. Check the instrumentation before believing the symptom.
4. Separate the trigger from the cause; the deploy is rarely the bug.

Follow-up

• What would you look at first, and what would it rule out?
• How would you tell a cause from a coincidence here?

Suggested practice checklist

Built from the rounds and topics Imagry candidates report.

Day 1 — Map the Imagry loop

• Write out the reported sequence: Introductory Call, Technical Assessments, In-Person Visits, Leadership and HR Alignment.
• For each round, write one sentence on what it is judging, from the description above, and mark the one you are least ready for.

Deliverable: A one-page map of the 4 reported rounds, with the weakest marked.

Day 2 — Work C++ (Programming)

• Spend the session on C++ (Programming), which Imagry candidates report being tested on.
• Write one worked example in C++ (Programming) and time yourself on it.

Deliverable: One timed worked example in C++ (Programming).

Day 3 — Work Deep Learning

• Spend the session on Deep Learning, which Imagry candidates report being tested on.
• Write one worked example in Deep Learning and time yourself on it.

Deliverable: One timed worked example in Deep Learning.

Day 4 — Work Motion Planning

• Spend the session on Motion Planning, which Imagry candidates report being tested on.
• Write one worked example in Motion Planning and time yourself on it.

Deliverable: One timed worked example in Motion Planning.

Day 5 — Answer out loud: C++ & Systems Programming

• Answer aloud, timed: Explain the difference between std::unique_ptr and std::shared_ptr and describe how they impact memory overhead in a multithreaded application.
• Answer aloud, timed: How does the virtual table (vtable) work in C++, and what is its performance cost during runtime?

Deliverable: Spoken answers to 2 reported C++ & Systems Programming question(s), under time.

Day 6 — Answer out loud: Deep Learning & Computer Vision

• Answer aloud, timed: How do you address the challenge of object occlusion in real-time 3D object detection?
• Answer aloud, timed: Describe the architectural differences between single-stage and two-stage object detectors, and how you choose between them for edge deployment.

Deliverable: Spoken answers to 2 reported Deep Learning & Computer Vision question(s), under time.

Day 7 — Answer out loud: Motion Planning & Robotics Algorithms

• Answer aloud, timed: Explain the difference between the A* algorithm and Dijkstra's algorithm. In what scenarios would you use a state-lattice planner instead?
• Answer aloud, timed: How do you formulate a trajectory optimization problem using Model Predictive Control (MPC) under dynamic vehicle constraints?

Deliverable: Spoken answers to 2 reported Motion Planning & Robotics Algorithms question(s), under time.

Behavioural preparation

Behavioural rounds judge the decision you made and what it cost.

How do you handle a situation where your technical proposal is rejected by a senior team member or architect?

Medium · Behavioral & Professional Alignment

Approach

1. Pick a story where you made the decision, not one where you watched it.
2. State the situation in two sentences and spend the rest on the reasoning.
3. Give the blast radius: what could have broken, and what you measured.
4. Name the disagreement and how you resolved it with evidence.

Follow-up

• What would you do differently if you ran that again?
• How did you know your change caused the improvement?

Give an example of how you balance the need for rapid prototyping with the strict quality standards of safety-

Medium · Behavioral & Professional Alignment

Give an example of how you balance the need for rapid prototyping with the strict quality standards of safety-critical software.

Approach

1. Pick a story where you made the decision, not one where you watched it.
2. State the situation in two sentences and spend the rest on the reasoning.
3. Give the blast radius: what could have broken, and what you measured.
4. Name the disagreement and how you resolved it with evidence.

Follow-up

• What would you do differently if you ran that again?
• How did you know your change caused the improvement?

Why are you interested in working on mapless autonomous driving technology compared to traditional map-based a

Medium · Behavioral & Professional Alignment

Why are you interested in working on mapless autonomous driving technology compared to traditional map-based approaches?

Approach

1. Pick a story where you made the decision, not one where you watched it.
2. State the situation in two sentences and spend the rest on the reasoning.
3. Give the blast radius: what could have broken, and what you measured.
4. Name the disagreement and how you resolved it with evidence.

Follow-up

• What would you do differently if you ran that again?
• How did you know your change caused the improvement?