A Software Engineer at Cleric works at the absolute frontier of AI engineering and infrastructure. Cleric is building the world's first autonomous AI agent designed to investigate and resolve complex production incidents. Instead of simply alerting human operators, this agent combines large language models (LLMs) with sophisticated tooling to navigate distributed systems, reason through ambiguous outages, and execute real-time corrective actions. As a Software Engineer on this team, your work has a direct impact on how modern tech companies manage reliability. You will design and implement the runtime systems, agent reasoning loops, and integration workflows that allow the agent to operate safely in high-scale production environments across fintech, ride-sharing, and autonomous vehicle sectors. This is not a standard backend role; you will be building the fundamental infrastructure that allows AI agents to interact with real-world, messy production environments. This position is ideal for engineers who possess strong software fundamentals but are also eager to solve the unique challenges of building reliable, deterministic systems on top of probabilistic AI models. You will be working in a fast-paced, high-impact environment where your contributions directly shape the core product and influence how engineering teams handle production failures globally.
Loom Video Submission
reportedSubmit a 5-minute Loom video showcasing a production incident story, focusing on context, symptom, root cause, and resolution.
What to demonstrate
- Submit a 5-minute Loom video showcasing a production incident story
- Focusing on context, symptom, root cause, and resolution
How to prepare
- Answer aloud and timed: Tell me about a time something broke in production. What was the root cause, and how did you investigate it?
- Answer aloud and timed: Describe a system you worked on that was exceptionally painful to deploy, debug, or operate. What made it so difficult?
Technical Screen
reportedParticipate in a standard technical screen to assess hands-on technical skills.
What to demonstrate
- Participate in a standard technical screen to assess hands-on technical skills
- Depth in Python
How to prepare
- Answer aloud and timed: If you could rebuild that painful system today with your current knowledge, what architectural changes would you make?
- Answer aloud and timed: Walk me through a 30-second demonstration of a command-line tool or utility you use daily. What does it do, and when is it most useful?
Deep-Dive Engineering Sessions
reportedEngage in interactive technical sessions that evaluate engineering capabilities and problem-solving skills.
What to demonstrate
- Engage in interactive technical sessions that evaluate engineering capabilities and problem-solving skills
- Depth in Python
How to prepare
- Answer aloud and timed: How do you balance speed of deployment with system stability when shipping updates to a live production environment?
- Answer aloud and timed: Design and implement a rate-limiting middleware for an API gateway. How do you handle distributed state?
Architectural and Behavioral Evaluation
reportedUndergo a final evaluation focusing on architectural thinking and behavioral fit within the team.
What to demonstrate
- Undergo a final evaluation focusing on architectural thinking and behavioral fit within the team
- Depth in Python
How to prepare
- Prepare three examples from your own work, each with a decision you made and an outcome you can quantify.
- Re-read the description of the architectural and behavioral evaluation above and write down what you would ask to confirm before it.
PracHub editorial advice for the preparation topics above.
Showcase SRE Empathy
Because Cleric's core product is an autonomous incident responder, always look at engineering problems through the lens of an SRE. Ask yourself: How would I monitor this? How would I debug this if it failed silently? How do we prevent this from waking up an engineer at 3 AM?
Embrace Radical Candor
During technical discussions and behavioral rounds, be direct, honest, and open about what you don't know. If your design has a weakness, point it out before the interviewer does. The team values self-awareness and direct communication far more than defensive positioning.
If you are asked to design a tool or integration, always address rate limits, token usage, and API failures
Showing that you think about these operational costs and constraints demonstrates a high level of engineering maturity.
Choose a category, try a prompt, then open its approach, worked solution or follow-up when you need it.
Implement a parser that extracts structured metrics from unstructured system logs in real-time.
Implement a parser that extracts structured metrics from unstructured system logs in real-time.
Approach
- Restate the input: its shape, its size, and what is guaranteed about it.
- Name the brute-force solution and its complexity before improving on it.
- Choose the data structure from the access pattern, not from familiarity.
- 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?
Diff a projection against the primary without per-row point reads
The listing projection has drifted and some rows show a stale version. The primary holds 40,000,000 resource rows across 12,000 tenants while serving 1,200 writes and 14,000 reads per second. The obvious repair, reading each resource row and comparing its version against the projection, is correct and would eventually finish. Explain precisely why it is unacceptable here, then give a diff that finds the differing rows, state its complexity, and make it safe to run against a live primary. Replication lag is usually under 100 ms and is not bounded.
Approach
- Quantify the naive cost rather than calling it slow: 40,000,000 point reads at even 0.5 ms each is over five hours serialised, and the only lever is concurrency, which is exactly what you cannot spend. The primary's pool is sized for the write path, and 40,000,000 random reads evict the buffer cache that sustains the 85 percent cache hit rate, so the audit degrades the system it is auditing.
- Replace random access with one ordered pass per side. Both sides can be read in (tenant_id, resource_id) order, which is a sequential scan on each and a merge join in O(n) time and O(1) memory. For a dense diff that is the whole answer, and it reads the primary once instead of 40,000,000 times.
- For the expected sparse case, compare range hashes instead of rows: partition the key space, compute per range an order-independent aggregate over hash(resource_id, version), compare aggregates, and descend only into ranges that differ. With d differing rows and branching factor B, at most d ranges mismatch per level, so the drill-down examines O(d log_B(n/d)) ranges and reads full rows only in mismatching leaves.
- Aggregate with a sum modulo 2^64 or a multiset hash, never XOR. XOR is order-independent but self-cancelling, so two rows wrong in the same way, or a row duplicated on one side, leave the range aggregate matching and the range is declared clean.
Follow-up
- The diff reports 900 stale rows. How do you decide between patching those rows and rebuilding the projection from resource_revision?
- Same job, but the projection lives in a search index that cannot be scanned in key order. What changes?
Merge partitioned event streams into one ordered feed with bounded lateness
The read-model service consumes 64 log partitions carrying about 4,000 events per second in total. Each partition is ordered within itself, but partitions drift by up to 30 seconds, and the activity feed must present a tenant's events in occurred_at order. Produce the merge. State its complexity, the buffer it requires in events and in bytes, what happens when one partition is idle, and what you do with an event that arrives after you have already emitted its position. Payloads average 1 KB.
Approach
- Merge with a min-heap over the 64 partition heads keyed on (occurred_at, event_id): O(log P) per event and O(n log P) overall. The tie-break on event_id is what makes the output deterministic when two partitions carry the same millisecond, which matters because the feed is paginated and a non-deterministic order reorders pages under the reader.
- Emitting the heap head is only correct once every partition has produced everything up to that timestamp, so the emit condition is a watermark: the minimum across partitions of the highest occurred_at seen, less the allowed lateness. Events are held until the watermark passes them, which is what turns individually ordered streams into a jointly ordered one.
- Size the buffer from the lateness rather than guessing: 4,000 events per second times 30 seconds is 120,000 buffered events, and at 1 KB each about 120 MB of heap. That number is the real price of the ordering guarantee and belongs in front of whoever asked for it.
- Handle the idle partition explicitly, because it fails the feed rather than corrupting it: a partition with no traffic never advances its own maximum, so the watermark freezes and output stops entirely. Either every partition emits a periodic idle marker carrying the broker's current time, or the watermark falls back to wall clock for a partition silent beyond a threshold.
Follow-up
- The lateness budget is raised to five minutes. What is the new buffer, and what besides memory changes?
- The consumer restarts. Where does it resume from, and what does the feed look like for the first 30 seconds?
Replace offset paging on the resource feed with keyset
resource holds resource_id, tenant_id, owner_user_id, title, body_ref, version, status ('draft','active','archived','deleted'), created_at, updated_at, deleted_at, with an index on (tenant_id, status, updated_at DESC, resource_id DESC). The listing endpoint returns active resources for one tenant, newest update first, 50 per page, today with LIMIT 50 OFFSET n. Tenants reach page 400 and rows are created while they read. Write the keyset query, define what the cursor carries and how it is encoded, and say which part of the index each predicate uses. Assume PostgreSQL 16.
Approach
- Name the two failures separately. OFFSET 20000 makes the server produce and discard 20,000 rows, so page cost grows with depth rather than with page size. Independently, any write that changes how many rows sort above the offset moves the window between two fetches, and the direction decides which anomaly you get: an insert lands at the head of updated_at DESC and pushes already-returned rows down past the boundary, so they are returned a second time; a delete above the offset, or a row whose updated_at is bumped above the cursor, pulls rows up and one is never returned at all. Nothing in the response reveals either.
- Write the seek: WHERE tenant_id = $1 AND status = 'active' AND (updated_at, resource_id) < ($2, $3) ORDER BY updated_at DESC, resource_id DESC LIMIT 50. The row-value comparison is one index range rather than a disjunction, and both columns are NOT NULL, which is what makes that comparison well defined.
- Map each predicate onto the index: tenant_id and status are equality on the leading columns, (updated_at, resource_id) is the range, and the ORDER BY matches the index order so no Sort node appears and the scan stops after 50 rows. The DESC in the definition only matters for mixed directions — a plain ascending btree on the same columns is read backwards for this query.
- Put both sort columns in the cursor and nothing the client can tamper with into another tenant: base64 of (updated_at, resource_id), validated server-side, with tenant_id taken from the principal.
Follow-up
- The client asks for 'jump to page 400'. What do you offer instead, and what does the honest version cost?
- Sort order becomes user-selectable across four columns. How many indexes is that, and which would you refuse to add?
Keep soft-deleted accounts from blocking re-registration
app_user holds user_id, tenant_id, email CITEXT, password_hash (NULL for SSO principals), email_verified_at, auth_version, status ('invited','active','suspended','deactivated'), created_at, updated_at, deleted_at. Two live accounts for one address inside a tenant must be impossible, but an address freed by a soft delete must be reusable, and the same tenant may delete and re-register it repeatedly. Write the uniqueness DDL for PostgreSQL 16, then the equivalent for MySQL 8 where partial indexes do not exist, and say what each permits once three deleted rows already hold that address.
Approach
- Start from what is actually unique: not (tenant_id, email), but (tenant_id, email) among live rows. PostgreSQL says that directly — CREATE UNIQUE INDEX app_user_live_email ON app_user (tenant_id, email) WHERE deleted_at IS NULL. A full constraint over the same two columns burns the address permanently the first time someone deletes an account.
- Keep case-insensitivity in the type or the index, never in the application: CITEXT as given, or UNIQUE (tenant_id, lower(email)) as an expression index where the extension is unavailable. A case-sensitive unique column is exactly how two accounts for one human appear.
- For MySQL 8 the predicate has to move inside the key: add a discriminator column that is a constant 0 while the row is live and is set to user_id on delete, with UNIQUE (tenant_id, email, deleted_marker). Live rows share the constant and still collide; deleted rows differ from each other and stop colliding.
- State the NULL variant and its dependency: leaving the marker NULL for deleted rows also works, because a unique index treats NULLs as distinct — true in MySQL, and true in PostgreSQL only under the default NULLS DISTINCT, which PostgreSQL 15 lets you reverse. Check the polarity against the three existing deleted rows: constant-on-live is what preserves the collision you want, and reversing it silently admits duplicate live accounts.
Follow-up
- A deleted account re-registers with the same address the next day. Do the old resource rows follow the new user_id, and how does the API keep the two principals apart?
- How do you honour an erasure request while resource_revision.actor_user_id still references this table?
If you could rebuild that painful system today with your current knowledge, what architectural changes would y
If you could rebuild that painful system today with your current knowledge, what architectural changes would you make?
Approach
- Fix the scope first: who calls this, how often, and what they do when it fails.
- Name the read and write paths separately; they rarely have the same bottleneck.
- Choose a partition key and say what query it makes expensive.
- 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?
Design and implement a rate-limiting middleware for an API gateway. How do you handle distributed state?
Design and implement a rate-limiting middleware for an API gateway. How do you handle distributed state?
Approach
- Fix the scope first: who calls this, how often, and what they do when it fails.
- Name the read and write paths separately; they rarely have the same bottleneck.
- Choose a partition key and say what query it makes expensive.
- 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 design a robust error-handling wrapper for an external API integration that frequently experienc
How would you design a robust error-handling wrapper for an external API integration that frequently experiences transient failures?
Approach
- Say who the caller is and what they do when the call fails halfway.
- Define the identity of a request so a retry cannot double-apply it.
- Separate accepted, pending, failed and confirmed; they are different facts.
- Design the error taxonomy before the success shape; callers branch on it.
Follow-up
- What happens if the caller retries after a timeout?
- How does a client discover it is on an old version of this contract?
How would you architect a secure sandboxing environment that allows an autonomous agent to safely execute shel
How would you architect a secure sandboxing environment that allows an autonomous agent to safely execute shell commands and database queries?
Approach
- Fix the scope first: who calls this, how often, and what they do when it fails.
- Name the read and write paths separately; they rarely have the same bottleneck.
- Choose a partition key and say what query it makes expensive.
- 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?
Design a real-time observability platform that logs every reasoning step, tool execution, and API call made by
Design a real-time observability platform that logs every reasoning step, tool execution, and API call made by an autonomous agent for human auditing.
Approach
- Fix the scope first: who calls this, how often, and what they do when it fails.
- Name the read and write paths separately; they rarely have the same bottleneck.
- Choose a partition key and say what query it makes expensive.
- 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 system you worked on that was exceptionally painful to deploy, debug, or operate. What made it so d
Describe a system you worked on that was exceptionally painful to deploy, debug, or operate. What made it so difficult?
Approach
- Establish what changed and when, before forming any theory.
- Pick a bisection that eliminates candidates whichever way it turns out.
- Check the instrumentation before believing the symptom.
- 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?
Walk me through a 30-second demonstration of a command-line tool or utility you use daily. What does it do, an
Walk me through a 30-second demonstration of a command-line tool or utility you use daily. What does it do, and when is it most useful?
Approach
- Establish what changed and when, before forming any theory.
- Pick a bisection that eliminates candidates whichever way it turns out.
- Check the instrumentation before believing the symptom.
- 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?
How do you balance speed of deployment with system stability when shipping updates to a live production enviro
How do you balance speed of deployment with system stability when shipping updates to a live production environment?
Approach
- Establish what changed and when, before forming any theory.
- Pick a bisection that eliminates candidates whichever way it turns out.
- Check the instrumentation before believing the symptom.
- 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?
Design an evaluation framework that systematically tests an AI agent's performance and safety across hundreds
Design an evaluation framework that systematically tests an AI agent's performance and safety across hundreds of simulated production incidents.
Approach
- Establish what changed and when, before forming any theory.
- Pick a bisection that eliminates candidates whichever way it turns out.
- Check the instrumentation before believing the symptom.
- 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?
Built from the rounds and topics Cleric candidates report.
Prepare, practise & reflect
One practical outcome each day. Spend longer where you need it.
0 / 7 done01Map the Cleric loop
- Write out the reported sequence: Loom Video Submission, Technical Screen, Deep-Dive Engineering Sessions, Architectural and Behavioral Evaluation.
- 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.
02Work Python
- Spend the session on Python, which Cleric candidates report being tested on.
- Write one worked example in Python and time yourself on it.
Deliverable: One timed worked example in Python.
03Work AI Agentic Systems
- Spend the session on AI Agentic Systems, which Cleric candidates report being tested on.
- Write one worked example in AI Agentic Systems and time yourself on it.
Deliverable: One timed worked example in AI Agentic Systems.
04Work System Design
- Spend the session on System Design, which Cleric candidates report being tested on.
- Write one worked example in System Design and time yourself on it.
Deliverable: One timed worked example in System Design.
05Answer out loud: Incident Management & Production Operations
- Answer aloud, timed: Tell me about a time something broke in production. What was the root cause, and how did you investigate it?
- Answer aloud, timed: Describe a system you worked on that was exceptionally painful to deploy, debug, or operate. What made it so difficult?
Deliverable: Spoken answers to 2 reported Incident Management & Production Operations question(s), under time.
06Answer out loud: Coding & Backend Implementation
- Answer aloud, timed: Design and implement a rate-limiting middleware for an API gateway. How do you handle distributed state?
- Answer aloud, timed: Write a program that coordinates and executes a dependency graph of asynchronous tasks. How do you handle task failures and retries?
Deliverable: Spoken answers to 2 reported Coding & Backend Implementation question(s), under time.
07Answer out loud: System Design & Agentic Architectures
- Answer aloud, timed: Design an evaluation framework that systematically tests an AI agent's performance and safety across hundreds of simulated production incidents.
- Answer aloud, timed: How would you architect a secure sandboxing environment that allows an autonomous agent to safely execute shell commands and database queries?
Deliverable: Spoken answers to 2 reported System Design & Agentic Architectures question(s), under time.
Expand any day for tasks and deliverables. Your progress is saved on this device.
Behavioural rounds judge the decision you made and what it cost.
Tell me about a time something broke in production. What was the root cause, and how did you investigate it?
Tell me about a time something broke in production. What was the root cause, and how did you investigate it?
Approach
- Pick a story where you made the decision, not one where you watched it.
- State the situation in two sentences and spend the rest on the reasoning.
- Give the blast radius: what could have broken, and what you measured.
- 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?
Write a program that coordinates and executes a dependency graph of asynchronous tasks. How do you handle task
Write a program that coordinates and executes a dependency graph of asynchronous tasks. How do you handle task failures and retries?
Approach
- Pick a story where you made the decision, not one where you watched it.
- State the situation in two sentences and spend the rest on the reasoning.
- Give the blast radius: what could have broken, and what you measured.
- 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?
How do you handle state management and context window limitations when an AI agent is executing a long-running
How do you handle state management and context window limitations when an AI agent is executing a long-running investigation over several hours?
Approach
- Pick a story where you made the decision, not one where you watched it.
- State the situation in two sentences and spend the rest on the reasoning.
- Give the blast radius: what could have broken, and what you measured.
- 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?
- 01
Tell me about a time something broke in production. What was the root cause, and how did you investigate it?
- 02
Write a program that coordinates and executes a dependency graph of asynchronous tasks. How do you handle task failures and retries?
- 03
How do you handle state management and context window limitations when an AI agent is executing a long-running investigation over several hours?
Do I need a background in Machine Learning or AI to apply?
No. While an interest in AI and agentic systems is required, this is primarily a backend and systems engineering role. The core challenges at Cleric involve building the robust infrastructure, integrations, and runtime environments that allow AI models to interact with the physical world safely and reliably.
Cleric Software Engineer candidate reports ↗What is the working style at Cleric?
Cleric operates with an in-person collaboration model in their San Francisco office. They believe that high-bandwidth, face-to-face communication is critical for moving fast and solving complex, open-ended problems. The team embraces a culture of radical candor, where feedback is direct, constructive, and aimed at building the best possible product.
Cleric Software Engineer candidate reports ↗How should I prepare for the 30-second tool demo in the Loom video?
Choose a tool you actually use daily and know intimately. It does not need to be highly obscure; it just needs to show that you care about developer productivity and command-line efficiency. Examples include a custom `git` alias, a `jq` command for parsing JSON, or an `fzf` configuration for fast terminal navigation. Explain what it does, show it in action, and explain why it saves you time.
Cleric Software Engineer candidate reports ↗What is the "Bar Raiser" round?
The Bar Raiser round focuses on your engineering practices, product mindset, and cultural alignment. Interviewers will explore how you handle feedback, your experience working in collaborative teams, and how you think about product quality. They want to ensure you align with Cleric’s values of high impact, rapid iteration, and radical candor.
Cleric Software Engineer candidate reports ↗What topics does Cleric test in interviews?
Cleric interviews most often cover Python, Backend Engineering, System Design, Integrations with Production Systems, and Tool Execution Workflows. The exact emphasis depends on the specific role you apply for.
Cleric Software Engineer candidate reports ↗Sources & methodology 3 sources ↗
No official company page is cited. Rounds and questions come from candidate reports and PracHub editorial material; each source shows the date it was read.
- 01Cleric Software Engineer candidate reports ↗
Company-reported rounds, questions and FAQ.
Candidate reports · Accessed 2026-09-22 - 02PracHub Software Engineer practice ↗
PracHub practice material, not company-reported.
PracHub page · Accessed 2026-09-22 - 03PracHub preparation framework ↗
PracHub preparation guidance.
PracHub page · Accessed 2026-09-22
Explore related engineering environments and compare the preparation focus.
Discord Software Engineer Interview Guide
Database practice for Discord candidates: how an index changes a query plan, and which anomalies survive a given isolation level, since read committed still permits non-repeatable reads.
CoreWeave Software Engineer Interview Guide
Failure-handling practice for CoreWeave software engineer candidates: requests whose outcome a timeout leaves unknown, where a retry is safe, and deduplicating at the receiver instead of assuming exactly-once delivery.
Rystad Energy Software Engineer Interview Guide
Prepare for Rystad Energy software interviews with worked Python validation, OOP and decorator examples, plus an ingestion case and practical checks.
Zoom Video Communications Software Engineer Interview Guide
What Zoom Video Communications asks Software Engineer candidates, from reported rounds and questions.