Interview guides · Astir it solutions

Astir it solutions Software Engineer Interview Guide

Astir it solutions Software Engineer interview guide: the two reported rounds, DBMS, OS and coding questions, and behavioral stories to prepare.

The 60-second brief

Astir it solutions is a technology services and consulting firm, so a Software Engineer there is placed on client engagements or internal projects rather than one fixed product. Candidates describe work across the whole delivery cycle: gathering requirements, designing and coding, testing and deployment, plus code reviews, technical documentation and production troubleshooting. Placements vary, from modernising older systems to building cloud-based services, which is why the interviews lean on portable fundamentals (databases, operating systems, data structures) and on how well you know one main language, rather than on any single client's stack.

This guide is for engineers interviewing for the Software Engineer role at Astir it solutions, from junior candidates to those with a few years of experience; the requirements candidates describe ask for a CS or related degree and prefer 2+ years of hands-on development. It covers the two reported stages, conceptual questions on DBMS, OS and data structures, basic coding and SQL, language mechanics in Java, Python, C# or C++, and the behavioral and motivation questions that come up when you are being considered for client placement.

Candidates report two stages over roughly 2-4 weeks: a combined technical and behavioral phone screen of up to 45 minutes, and an HR and behavioral evaluation that is often held the same day.

Focus areas

Basic Coding / Programming
Data Structures
DBMS (Database Management Systems)

Overview

The questions candidates report are short and conceptual: explain ACID, define a deadlock and how to prevent it, describe how a hash map handles collisions, contrast primary and foreign keys, lay out the memory of a C/C++ program. None of them needs a whiteboard design. What separates answers is precision: a one-sentence definition, the mechanism underneath it, and one concrete example, delivered without wandering. Rehearse each topic until you can give that three-part answer out loud.

The coding portion candidates describe is basic: the second largest element in an array, the second highest salary in SQL, reversing a string without a built-in, a short inheritance snippet. These are easy to get roughly right and easy to get subtly wrong. The preparation value is in edge cases: duplicates, arrays with fewer than two distinct values, negative numbers, NULL salaries and ties. Write each one by hand, then test it against those inputs.

Candidates report choosing their own language and then fielding follow-ups on how it works underneath: garbage collection, exception handling, inheritance, and memory differences between Java and C++. Pick the language you can defend at that depth, and remove anything from your resume you could not explain at the level of the runtime.

Because the role places you on client work, the behavioral and motivation questions ask why consulting suits you, where you want to be in 3 to 5 years, what you would contribute, and how you have handled challenges, missed deadlines and colleagues who work differently. Prepare five or six specific stories and short, honest motivation answers that address the trade-offs of client placement, and bring your own questions about project assignment, workload and contract terms to the HR stage.

The Astir it solutions interview process

Candidates describe a short loop: a phone screen that mixes technical and background questions, followed by an HR and behavioral evaluation, often on the same day.

Combined Technical and Behavioral Screening

Candidates report this stage as a phone interview of up to 45 minutes that covers your background, your future plans and fundamental technical knowledge. They describe it as a run of quick conceptual questions rather than a request to build a system from scratch. Treat it as a rapid oral exam: answer each question with a crisp definition, the mechanism behind it and one example, then stop and let the next question come. Expect the coding and SQL items to be small enough to talk through or write quickly, so accuracy on edge cases matters more than cleverness.

What to demonstrate

Fundamental technical knowledge across DBMS, operating systems and data structures
Command of programming concepts in the language you choose
A clear account of your background and future plans
Writing small pieces of working code and SQL correctly

How to prepare

Write a three-line answer (definition, mechanism, example) for ACID, deadlock prevention, hash map collision handling, primary versus foreign keys, and the C/C++ memory layout, then say each aloud.
Code the second-largest-element function and the second-highest-salary query by hand, and test them on duplicates, single-value input, negatives and NULLs.
Choose your primary language and rehearse its garbage collection or memory model, exception handling and inheritance with a short snippet for each.
Prepare a 60-second walk through your background that ends with where you want your career to go next.

HR and Behavioral Evaluations

Candidates report that the HR and behavioral evaluation is often held on the same day as the technical screen, and that it looks at interpersonal skills and fit. The reported behavioral questions cover major challenges, collaborating with a different working style, missed deadlines, projects you are proud of and achievements, and the motivation questions cover how you heard about the company, your 3-to-5-year plans, why consulting, and the environment you work best in. Answer stories in STAR form with your own actions in the middle, and use this stage to ask about the client project you might join, the workload, contract terms and benefits.

What to demonstrate

Interpersonal skills and how you work with others
Fit with a consulting role that places you on client projects
Career motivation and what you would contribute

How to prepare

Write five STAR stories mapped to the reported prompts: a major challenge, a different working style, a missed deadline, a project you are proud of, and an achievement or award.
Write short answers to why consulting, your 3-to-5-year plan, how you heard about the company and your preferred work environment, each naming one trade-off you accept.
List the questions you will ask: which client project, expected workload, contract terms, benefits and compensation structure.
Rehearse both stages back to back so you can switch from technical answers to stories in a single sitting.

Where candidates lose points

Turning a quick conceptual question into a five-minute lecture, or giving a vague one-liner with no mechanism.

Candidates describe rapid, short questions. Rehearse each fundamentals topic (ACID, deadlock, hash map collisions, keys, memory layout) as definition, mechanism, example, and stop there. If more depth is wanted, the next question will ask for it.

Claiming a language or tool on your resume and then being unable to explain how it works underneath.

Candidates report follow-ups on the language they choose. If you list Java, be ready for generational GC, checked exceptions and inheritance rules; if C++, for the memory layout, RAII and virtual destructors. Cut anything you cannot defend at that level.

Basic coding answers that fail on edge cases: a second-largest function initialised to 0 that breaks on negatives, or a salary query that returns the top salary again when it is duplicated.

For every small coding or SQL question, ask about duplicates and empty input first, initialise to the minimum value or null, use DENSE_RANK or a strictly-less-than MAX subquery for ties, and handle NULL salaries explicitly.

Answering the deadlock question with only 'use locks carefully', or mixing up prevention, avoidance and detection.

Learn the four Coffman conditions and which technique breaks which one: lock ordering, all-at-once acquisition, tryLock with timeout, the Banker's algorithm, and wait-for graph detection with victim rollback.

Behavioral and motivation answers that ignore the consulting context: stories narrated in the plural, and a why-consulting answer with no trade-offs.

Put your own actions at the centre of every STAR story, and prepare motivation answers that admit the costs of client placement (context switching, legacy work, changing teams) and explain why you accept them.

Questions & practice

How does a hash map resolve collisions under the hood?

Medium · Technical Fundamentals

Approach

Start with indexing: the key's hash is reduced to a bucket index, for example hash mod capacity, or a bitmask when the capacity is a power of two. A collision is two keys landing in the same bucket.
Separate chaining stores a list per bucket and compares keys with equals; Java's HashMap converts a long chain into a red-black tree, so a bad bucket costs O(log n) instead of O(n).
Open addressing stores entries in the array itself and probes for the next slot: linear probing, quadratic probing or double hashing. Deletion needs tombstones so lookups do not stop early. Python's dict uses open addressing.
Explain the load factor and resizing: when entries divided by capacity passes a threshold (0.75 in Java's HashMap), the table grows and every entry is rehashed, which keeps average operations O(1) amortised.
State the complexity and the contract: O(1) average, worse when many keys collide, and keys must keep equals and hashCode consistent and must not change while stored.

Follow-up

What happens if you mutate a key object after inserting it into a hash map?
Why do many implementations keep the capacity a power of two?
How does lookup in a hash map compare with a balanced binary search tree, and when would you choose the tree?

Describe the memory layout of a C/C++ program.

Medium · Technical Fundamentals

Approach

Walk the segments from low to high addresses: text (machine code, read-only), read-only data such as string literals, initialised data (.data), uninitialised data (.bss, zero-filled at load), the heap, and the stack near the top, with arguments and environment above it.
Say what grows where: the heap grows upward through malloc or new; the stack grows downward with one frame per call holding locals, saved registers and the return address. Shared libraries and mmap regions sit between them.
Classify a snippet: a global int set to 5 goes in .data, an uninitialised static goes in .bss, a local pointer lives on the stack while the block it points to from malloc lives on the heap, and a static local lives in .data or .bss, not on the stack.
Tie the layout to bugs: deep recursion overflows the stack, a missing free leaks heap memory, returning a pointer to a local leaves it dangling, and writing to a string literal is undefined behaviour.
Mention that these are virtual addresses per process; the OS maps them to physical pages.

Follow-up

Why is stack allocation cheaper than heap allocation?
Where does a const global variable end up, and why?
What happens when two threads run in the same process: which segments are shared and which are per thread?

Write a program to find the second largest element in an array.

Medium · Basic Coding and Language Mechanics

Approach

Clarify first: does a repeated maximum count, so that [5, 5, 3] returns 5 or 3? What should come back when there are fewer than two distinct values: an exception, null or a sentinel?
Make one pass keeping first and second. If x is greater than first, move first into second and set first to x; otherwise, if x differs from first and is greater than second, set second to x. This runs in O(n) time with O(1) space.
Initialise to the smallest possible value or to null, never to 0, so arrays of negative numbers work.
Contrast this with sorting and taking the second distinct value from the end: O(n log n) and it changes or copies the array, so the single pass is better.
Test aloud on [3], [7, 7, 7], [-4, -2, -9], [1, 2] and an array whose maximum appears twice.

Follow-up

How would you generalise to the k-th largest element (a min-heap of size k, or quickselect)?
What changes if the numbers arrive as a stream you cannot store?
Return the index of the second largest instead of its value: what about ties?

How does garbage collection work in Java (or your preferred language)?

Medium · Basic Coding and Language Mechanics

Approach

In Java, an object is garbage once it is unreachable from the GC roots: local variables on thread stacks, static fields, active threads and JNI references. Reachability, not reference counting, so cycles are collected.
Describe the generational heap: new objects go into Eden; a minor collection copies survivors between two survivor spaces and promotes long-lived objects to the old generation, which is collected less often with mark-sweep-compact.
Name the collectors: Serial, Parallel, G1 (the default since Java 9, region-based) and the low-pause ZGC and Shenandoah. Explain stop-the-world pauses and the throughput versus latency trade-off.
Show that Java can still leak: objects held by static collections, caches without eviction, unremoved listeners. System.gc() is only a hint, and finalize is deprecated in favour of try-with-resources.
If you pick another language, give its model: CPython uses reference counting plus a cycle collector; C++ has no GC and relies on RAII, unique_ptr and shared_ptr, with weak_ptr to break cycles.

Follow-up

How would you find a memory leak in a running Java service?
What are the memory management differences between Java and C++?
What are weak and soft references used for?

Explain the concept of inheritance and write a brief code snippet demonstrating it.

Medium · Basic Coding and Language Mechanics

Approach

Define it as an is-a relationship: a subclass inherits the fields and methods of a base class, adds its own and overrides behaviour, which lets code written against the base type work with any subclass.
Write a short snippet: an abstract Shape with an area() method, and Circle and Rectangle subclasses that call super() in their constructors and override area(). Then loop over a List of Shape and call area() to show runtime polymorphism through dynamic dispatch.
Cover the language specifics for your chosen language: Java has single class inheritance plus multiple interfaces; C++ needs virtual for dynamic dispatch and a virtual destructor when deleting through a base pointer; Python resolves multiple inheritance with the C3 method resolution order.
Name the types (single, multilevel, hierarchical, multiple) and the diamond problem.
Say when not to inherit: if the relation is has-a, use composition. Java's Stack extending Vector is the standard example of inheritance exposing methods that break the abstraction.

Follow-up

What is the difference between an abstract class and an interface?
How does C++ solve the diamond problem?
Explain overriding versus overloading with an example.

Write a basic SQL query to find the second highest salary from an employee table.

Medium · Basic Coding and Language Mechanics

Approach

Start with the portable form: SELECT MAX(salary) FROM employee WHERE salary < (SELECT MAX(salary) FROM employee). It returns NULL when there is no second distinct salary, which is usually the result wanted.
Offer the ranking version for ties: rank with DENSE_RANK() OVER (ORDER BY salary DESC) in a subquery and filter rank = 2. Explain why DENSE_RANK, not RANK or ROW_NUMBER: duplicates of the top salary must not consume rank 2.
Mention SELECT DISTINCT salary ... ORDER BY salary DESC LIMIT 1 OFFSET 1, and note that it returns an empty result rather than NULL unless wrapped in an outer SELECT.
Handle NULL salaries: MAX ignores them, but ORDER BY salary DESC places NULLs first in PostgreSQL, so add WHERE salary IS NOT NULL to the ranking and LIMIT forms.
Note that an index on salary lets the MAX forms read the top of the index instead of scanning the table.

Follow-up

Generalise to the N-th highest salary.
Return the second highest salary in each department.
List the names of every employee who earns the second highest salary.

Replace offset paging on the resource feed with keyset

Easy · keyset pagination · composite index · sort order

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?

Explain the ACID properties in a Database Management System.

Medium · Technical Fundamentals

Approach

Define each letter with its mechanism: Atomicity means a transaction commits entirely or not at all, implemented with an undo log or write-ahead log so a failure rolls back partial work.
Consistency means a transaction moves the database from one valid state to another: primary key, foreign key, unique, check constraints and application invariants all hold after commit.
Isolation means concurrent transactions behave as if run one at a time; name the SQL levels (read uncommitted, read committed, repeatable read, serializable) and the anomalies they allow: dirty reads, non-repeatable reads, phantoms.
Durability means a committed change survives a crash because the log record is flushed to disk before the commit is acknowledged.
Anchor it in one example: a transfer that debits account A and credits account B, and show what each property prevents if the server dies between the two updates.

Follow-up

Which anomaly does each isolation level permit, and which level does your usual database default to?
How does a write-ahead log provide both atomicity and durability?
How do ACID guarantees compare with the eventual consistency many NoSQL stores offer?

Can you explain the difference between a primary key and a foreign key?

Medium · Technical Fundamentals

Approach

A primary key uniquely identifies each row, cannot be NULL, and there is one per table, though it may span several columns. The database backs it with a unique index.
A foreign key is a column whose values must match a primary or unique key in another table (or the same table), enforcing referential integrity. It can repeat, which is how one-to-many relationships work, and can be NULL unless declared otherwise.
Use a concrete pair: orders.customer_id references customers.id. Many orders can point to one customer, and an order cannot reference a customer that does not exist.
Cover referential actions on the parent: ON DELETE CASCADE, SET NULL or RESTRICT, and say which you would choose for orders when a customer is deleted.
Add the practical point: index the foreign key column, because joins and parent deletes use it, and some databases such as PostgreSQL do not create that index automatically.

Follow-up

What is the difference between a primary key and a unique constraint?
Would you choose a surrogate key or a natural key for a customers table, and why?
How would you model a many-to-many relationship between students and courses?

How did you hear about Astir IT solutions?

Medium · Motivation and Career Goals

Approach

Give the honest channel: a job board, a referral, a recruiter message or a campus event.
Add what you learned afterwards: that Astir it solutions is a technology services and consulting firm placing engineers on client projects across industries, and why that model interested you.
Keep it short and do not invent knowledge of specific clients or projects you have not read about.
Bridge to motivation in one sentence, which sets up the why-consulting question.

Follow-up

What do you know about how the company works with clients?
What made you apply rather than ignore the posting?
Are you interviewing with other consulting firms?

What are your future plans for your career over the next 3 to 5 years?

Medium · Motivation and Career Goals

Approach

Give a concrete path: year one, become productive quickly in your main language and on your first client project; years two and three, broaden across domains and stacks; years three to five, lead the technical side of an engagement or mentor newer engineers.
Tie the plan to the role: varied client work is how you intend to build that breadth.
Name a skill you are actively building now, such as a cloud platform (AWS or Azure appear among the nice-to-have skills candidates describe).
Avoid plans that signal you will leave soon, and avoid vague answers like wanting to grow.

Follow-up

What skills are you missing for that path?
Would you prefer a technical lead or a management track?
How will you measure whether you are on track?

What specific contributions do you believe you can make to our company?

Medium · Motivation and Career Goals

Approach

Map two or three specific strengths to the work candidates describe: depth in one language, speed at learning a new stack, comfort talking to non-technical stakeholders, production troubleshooting.
Back each with a short proof from your history: a stack you learned for a project, a client or user you explained a trade-off to, an incident you resolved.
Be realistic about where you are; a junior candidate can offer fast ramp-up and reliable delivery rather than architecture leadership.
End with what you would contribute in your first months on a client project.

Follow-up

Give an example of learning a new technology quickly.
Which of your weaknesses would show on a client project?
How would you add value beyond your assigned tasks?

Why are you interested in a consulting-style engineering role?

Medium · Motivation and Career Goals

Approach

Give reasons tied to consulting itself: exposure to several industries and stacks, faster learning through variety, and working directly with the people who use the software.
Show you understand the trade-offs: switching context between engagements, client deadlines, being placed on a legacy system, ramping up repeatedly.
Support it with evidence: a time you joined an unfamiliar codebase or stack and became productive, or a time you worked with an external stakeholder.
Avoid answers that make consulting sound like a stopgap before a product company.

Follow-up

How would you feel about being placed on a legacy maintenance project?
How do you ramp up on an unfamiliar codebase in your first week?
How would you handle a client who changes requirements late?

What type of work environment allows you to do your best work?

Medium · Motivation and Career Goals

Approach

Describe conditions, not perks: clear goals, regular feedback, ownership of a piece of work, and teammates who review each other's code.
Show adaptability: say how you still deliver when those conditions are missing, because client environments differ.
Give an example of an environment where you did your best work and why.
Avoid rigid demands (fully remote only, no meetings) unless they are real requirements, in which case raise them with HR directly.

Follow-up

Describe an environment where you struggled and what you did about it.
How do you work when requirements are unclear?
How do you prefer to receive feedback?

What is a deadlock in an Operating System, and how can it be prevented?

Medium · Technical Fundamentals

Approach

Define it: a set of threads or processes each waiting for a resource another one holds, so none can proceed. Give the textbook case: thread 1 locks A then B, thread 2 locks B then A.
List the four Coffman conditions, all required: mutual exclusion, hold and wait, no preemption, circular wait.
Prevention breaks one condition: a global lock ordering removes circular wait; acquiring all resources at once removes hold and wait; tryLock with a timeout and release removes the no-preemption condition.
Separate prevention from avoidance (the Banker's algorithm grants a request only if the system stays in a safe state) and from detection and recovery (find a cycle in the wait-for graph, then abort or roll back a victim, which is what databases do).
Close by distinguishing deadlock from livelock (threads keep acting but make no progress) and starvation (one thread never gets the resource).

Follow-up

How does a relational database detect and resolve a deadlock between two transactions?
Walk through the Banker's algorithm on a small example.
If you saw a production service hang, how would you confirm it was a deadlock?

Suggested practice checklist

Each day maps to one cluster of the reported questions: databases, operating systems and memory, data structures and coding, language mechanics, behavioral stories and motivation, then a full same-day rehearsal of both reported stages.

Day 1 — Databases: ACID, keys and the salary query

Write definition, mechanism and example for each ACID property, plus a table of isolation levels and the anomalies each allows.
Sketch a customers and orders schema with primary and foreign keys and choose an ON DELETE action for each relationship.
Write the second-highest-salary query three ways (MAX subquery, DENSE_RANK, LIMIT/OFFSET) and test them on ties and NULL salaries.
Rewrite an OFFSET-paged query as keyset paging to practise reading a query through its index.

Deliverable: An ACID and isolation sheet, an annotated schema, three tested salary queries and one keyset-paging query.

Day 2 — Operating systems and memory

Write two threads that deadlock on two locks, then fix it with a global lock order and again with tryLock and timeout.
Write out the four Coffman conditions and match each prevention technique to the condition it breaks; add a short Banker's algorithm example.
Draw the C/C++ memory layout and label where a global, a static local, a local pointer, its malloc'd block and a string literal live.
Write a one-paragraph answer on process versus thread: what each shares and what each owns.

Deliverable: A deadlocking program with two fixes, a conditions table, a labelled memory diagram and a process-versus-thread answer.

Day 3 — Data structures and basic coding

Implement a small hash map with separate chaining and resizing at a load factor of 0.75, then explain how open addressing with tombstones would differ.
Code the second-largest function in your chosen language and run it on [3], [7, 7, 7], negatives and a duplicated maximum.
Write a string reversal without built-ins, and compare hash map and binary search tree lookup complexity in two sentences.

Deliverable: A working chained hash map, a tested second-largest function, a string reversal and a complexity comparison note.

Day 4 — Language mechanics in your chosen language

Write a page on how your language manages memory: Java's generational GC and collectors, CPython's reference counting plus cycle collector, or C++ RAII and smart pointers.
Write an inheritance snippet with an abstract base, two subclasses, an override and a polymorphic loop; note the diamond problem for your language.
Write an example of your error-handling approach: try-with-resources or RAII, a wrapped domain exception and a retried transient failure.

Deliverable: A memory-management page, an inheritance snippet and an error-handling example, each in your chosen language.

Day 5 — Behavioral stories

Write STAR stories for a major challenge, a different working style, a missed deadline, a project you are proud of and your top achievement.
For each, underline the sentences that describe your own actions and cut any that describe only the team.
Tell each story aloud and add one measurable result to any story that lacks one.

Deliverable: Five written STAR stories, each with personal actions and a measurable result, rehearsed aloud.

Day 6 — Motivation and the HR stage

Write short answers to how you heard about the company, your 3-to-5-year plan, the contribution you would make, why consulting and your preferred work environment.
Add one trade-off of client placement to the why-consulting answer and say why you accept it.
List your questions for HR: likely client project, workload, contract terms, benefits and compensation structure.

Deliverable: Five motivation answers and a list of questions for the HR stage.

Day 7 — Same-day mock of both stages

Have someone fire fundamentals questions at you from days 1-4 in random order and answer each in under a minute.
Immediately afterwards, run the behavioral and motivation set without a break, as candidates report both stages on the same day.
Note every answer that ran long or missed an edge case and rewrite it.

Deliverable: A recorded mock of both stages and a list of rewritten answers.

Behavioural preparation

Candidates report that the HR and behavioral evaluation covers interpersonal skills and fit, and that the phone screen also asks about your background and future plans. For a role placed on client projects, prepare stories that show you adapting to new teams, owning problems and communicating clearly when things slip.

Why are you interested in a consulting-style engineering role?

How do you handle exceptions and errors in your code?

Medium · Basic Coding and Language Mechanics

Approach

Treat this as a language mechanics question, then back it with a real example. In Java, explain checked versus unchecked exceptions, try/catch/finally, and try-with-resources with AutoCloseable so files and connections always close.
State the principles: catch an exception only where you can handle it; never swallow it silently; log it once with context at the boundary; wrap low-level exceptions in domain exceptions while keeping the cause.
Separate error kinds: validate input at the edges and fail fast on programmer errors, but retry transient failures such as timeouts with backoff, and only when the operation is idempotent.
Give the equivalents if you chose another language: C++ uses RAII so destructors clean up during stack unwinding, and noexcept marks functions that must not throw; Python uses try/except/else/finally and context managers.
Finish with one specific case from your own code, in first person: the failure, how it was caught, what the user saw and what was logged.

Follow-up

What happens if a finally block contains a return statement?
Why is throwing from a C++ destructor dangerous?
When would you return an error value instead of throwing?