Technical Interviewer’s Questions

1\ how do you approach api versioning, and what strategy would you recommend for a rapidly evolving service? great response "i approach api versioning by first considering the needs of api consumers for a rapidly evolving service, i'd recommend url based versioning (like /api/v1/resource ) because it's explicit and easy for clients to understand however, i also use accept headers for more granular control when needed i maintain backward compatibility for at least two previous versions and communicate deprecation timelines clearly to clients for breaking changes, i create comprehensive migration guides and provide tooling to help clients transition i've found that automated compatibility tests between versions are crucial to catch unintended breaking changes before they reach production " mediocre response "i usually put the version in the url like /api/v1/resource it makes it clear which version is being used when we need to make changes, we create a new version and let users know they should update we try to keep the old version working for a while before removing it " poor response "we just increment the version number whenever we need to make changes if users are still using the old version, they need to update their code i don't think it's worth the extra effort to maintain multiple versions simultaneously since it creates technical debt i usually just focus on the latest version and make sure it has all the features we need " 2\ explain how you would design a rate limiting system for an api great response "i'd implement a token bucket algorithm which allows for bursts of traffic but maintains an average rate limit i'd use redis to store counters with appropriate ttls matching our rate limit windows for distributed systems, i'd use redis's atomic operations to prevent race conditions the implementation would include tiered rate limits—different limits for authenticated vs unauthenticated requests, different user tiers, and different endpoints based on resource cost i'd ensure the system returns appropriate http 429 responses with retry after headers and clear documentation i'd also add monitoring to detect abuse patterns, and gradually escalate from rate limiting to temporary bans for persistent abusers " mediocre response "i'd implement a counter in redis that tracks requests by ip or api key within a time window when a request comes in, we increment the counter and check if it exceeds the limit if it does, we return a 429 too many requests response we'd have different limits for different types of users the counters would expire after the time window to reset the limits " poor response "i'd create a database table to track each user's requests with timestamps for each request, i'd count how many requests they've made in the last hour and block them if they've gone over the limit if performance becomes an issue, we could cache the counts or maybe add more database capacity the database approach gives us good tracking for auditing purposes " 3\ how would you design a system to handle and process a large number of asynchronous tasks? great response "i'd implement a message queue architecture using something like rabbitmq or apache kafka as the backbone tasks would be serialized and published to appropriate topic queues, with consumer services processing them asynchronously i'd design the system for horizontal scalability, allowing us to add more worker nodes during high load periods for reliability, i'd implement dead letter queues, retry mechanisms with exponential backoff, and idempotent operations to handle duplicate processing i'd also add comprehensive monitoring with metrics for queue depths, processing times, and error rates, with alerts for anomalies for task prioritization, i'd implement multiple queues with different priority levels and ensure higher priority tasks get processed first " mediocre response "i'd use a message queue like rabbitmq and set up worker processes to handle the tasks each task would get added to the queue and then processed when a worker is available if a task fails, we could retry it a few times before moving it to a dead letter queue we'd monitor the queue size to make sure it doesn't get too large " poor response "i'd create a database table to store tasks and their status we'd have a cron job that runs every few minutes to pick up pending tasks and process them each task would be marked as in progress while it's being worked on, and then complete or failed when finished if we need more throughput, we could run multiple instances of the cron job " 4\ what considerations would you make when designing a restful api? great response "i prioritize designing resource oriented endpoints that map to domain entities rather than operations i use proper http methods (get for retrieval, post for creation, etc ) and corresponding status codes for error handling, i implement consistent error response formats with useful messages and references i design with pagination, filtering, and sorting for collection endpoints using query parameters for performance and scalability, i implement caching with appropriate etag/if none match headers and cache control directives i also focus on security by implementing proper authentication, authorization, input validation, and rate limiting i document the api extensively using openapi/swagger specifications and provide examples for common operations finally, i ensure the api is evolvable by using hateoas principles where appropriate and designing with backward compatibility in mind " mediocre response "i make sure to use the correct http methods and status codes get requests should be idempotent and only for retrieving data i structure the api around resources like /users or /orders rather than actions i implement authentication using something like jwt tokens and make sure to validate all inputs i document the endpoints so other developers know how to use them " poor response "i focus on making the api functional and easy to use i create endpoints that match what the frontend needs, and use post requests since they can handle more complex data i make sure to return json data and appropriate success or error messages if the api gets slow, i'd add caching or optimize the database queries behind it " 5\ how do you approach database schema design, and what considerations do you make when choosing between normalization and denormalization? great response "i start database design by modeling the domain entities and their relationships, using techniques like entity relationship diagrams i consider access patterns early to inform the design for normalization decisions, i balance theoretical correctness with practical performance needs i typically start with a normalized design (3nf) to prevent data anomalies and ensure data integrity, then strategically denormalize based on query patterns and performance profiling frequent joins, complex aggregations, or high read scenarios might benefit from calculated fields or summary tables i use database specific features like materialized views for read heavy scenarios for scaling, i consider partitioning strategies early i also implement necessary constraints, indexes, and evaluate storage requirements in distributed systems, i'm especially careful about denormalization since eventual consistency can complicate updates across denormalized data " mediocre response "i try to normalize the database to around third normal form to avoid data duplication if we notice performance issues with certain queries, we might denormalize some tables by adding redundant data i make sure to add indexes on columns that are frequently used in where clauses or joins for large tables, i'd consider partitioning to improve query performance " poor response "i start by looking at what data we need to store and create tables for each main type of data i add foreign keys to connect related data if queries are slow, i add indexes to make them faster sometimes i duplicate data across tables if it's easier than doing complex joins i tend to design for what the application needs right now rather than overthinking future requirements that might not materialize " 6\ explain your approach to error handling and logging in a production backend system great response "i implement a multi layered approach to error handling at the code level, i differentiate between expected errors (like validation failures) and unexpected exceptions, handling them appropriately i create custom exception types for different error categories to enable specific handling for logging, i use structured logging with contextual information and consistent formats that facilitate analysis i implement different severity levels and ensure sensitive data is never logged in production systems, i aggregate logs centrally using tools like elk stack or grafana loki, with real time alerting for critical errors i also track error rates and patterns to identify systemic issues for user facing errors, i return appropriate http status codes with helpful messages without exposing internal details finally, i implement circuit breakers and fallback mechanisms for dependent services to prevent cascading failures " mediocre response "i use try catch blocks to handle exceptions and make sure they don't crash the application for logging, i use a library like winston or log4j to output logs with different severity levels in production, we send logs to a centralized system so we can search them when issues occur i make sure to log enough context to understand what happened but avoid logging sensitive information " poor response "i surround code with try catch blocks and log any errors that occur i usually log to files that we can check later if something goes wrong for users, i return error messages that help them understand what went wrong if third party services fail, i catch those exceptions and return appropriate error responses we have monitoring that alerts us if too many errors are happening " 7\ how do you ensure the security of a backend system? great response "i approach security holistically, implementing defense in depth for authentication, i use industry standards like oauth 2 0 with short lived jwt tokens, protecting against session hijacking i implement fine grained authorization checks at both controller and service layers for data protection, i encrypt sensitive data both in transit (tls 1 3) and at rest, using appropriate encryption algorithms and key management i practice strict input validation at all entry points, using parameterized queries to prevent sql injection and output encoding to prevent xss i implement rate limiting, csrf protection, and secure headers (content security policy, x content type options, etc ) i regularly update dependencies to address vulnerabilities and use automated scanning tools in the ci pipeline i also follow the principle of least privilege for all system components and maintain comprehensive audit logs of security relevant events finally, i perform regular security reviews and stay updated on owasp top 10 and other security best practices " mediocre response "i implement authentication using jwt tokens and make sure all endpoints require proper authentication i validate all user inputs to prevent sql injection and xss attacks i use https for all communications and make sure sensitive data is encrypted in the database i keep dependencies updated to avoid known vulnerabilities and implement csrf protection for browser clients " poor response "i make sure to use https and implement authentication with username and password i validate inputs on important endpoints and make sure to use prepared statements for database queries our security team handles most of the security concerns, so i focus on implementing the features they recommend we use a waf in production that blocks most common attack patterns " 8\ how do you approach testing in backend development? great response "i implement a comprehensive test pyramid with unit, integration, and end to end tests at the unit level, i focus on testing business logic with both positive and negative test cases, using mocks for external dependencies to ensure tests are fast and deterministic for integration tests, i test interactions between components, using test containers for dependencies like databases to ensure tests run in environments similar to production i implement contract tests for service boundaries, particularly for microservices for critical paths, i add end to end tests that verify the entire system works together i practice tdd for complex logic and use property based testing for data intensive operations i measure test coverage but focus on meaningful coverage rather than arbitrary percentages all tests run in ci/cd pipelines with appropriate parallelization i also implement chaos engineering practices in staging environments to test resilience and implement monitoring driven development with synthetic transactions in production " mediocre response "i write unit tests for business logic and service layers using a framework like junit or jest i mock external dependencies to keep tests focused and fast for integration testing, i test how components work together, often using an in memory database i make sure critical paths have good test coverage all tests run in our ci pipeline before deployment " poor response "i write tests for the most important functionality to make sure it works as expected i focus on testing the happy path since that's what most users will experience we have a qa team that does more extensive testing before releases if there's a bug, i add a test case to prevent regression i try to keep a reasonable test coverage percentage, but i prioritize delivering features on time " 9\ describe how you would design a caching strategy for a backend service great response "i design cache strategies based on data access patterns and consistency requirements i implement a multi level caching approach application level caching for frequently computed results, distributed caching with redis for shared data, and database query caching for expensive queries i use different cache invalidation strategies depending on the use case time based expiration for relatively static data, event based invalidation for data that changes on specific operations, and write through caching for data that needs immediate consistency i implement cache keys carefully to avoid collisions and enable efficient invalidation for hot data, i use read aside caching with appropriate ttls, while for write heavy workloads, i might use write behind caching to batch updates i monitor cache hit rates, memory usage, and eviction rates to continuously optimize the strategy for distributed systems, i implement consistent hashing to distribute cache load and handle node failures gracefully i'm also careful about the thundering herd problem, implementing techniques like request coalescing or staggered expiration times " mediocre response "i use a combination of memory caching and distributed caching with redis for frequently accessed data that doesn't change often, i cache it with an appropriate ttl for data that changes more frequently, i implement cache invalidation when the data is updated i make sure to use meaningful cache keys that include all relevant parameters i also implement monitoring to track cache hit rates and adjust the strategy if needed " poor response "i'd use redis to cache api responses and database query results we'd set expiration times based on how often the data changes when data is updated, we'd clear the related cache entries if the cache gets too big, we'd increase the redis capacity or reduce the ttl of less important items caching is pretty straightforward you just store the results and check the cache before doing expensive operations " 10\ how would you handle database migrations in a production environment? great response "i approach database migrations with a focus on zero downtime deployment i use a versioned migration system like flyway or liquibase that tracks applied changes and executes them in order i follow a forward only migration philosophy—all changes are additive, with backward compatibility maintained through multiple deployment phases for schema changes, i first deploy code that can work with both old and new schemas, then apply the migration, and finally deploy code that uses the new schema for large tables, i implement migrations that operate in small batches to avoid locking tables for extended periods before production, i test migrations thoroughly in staging environments with production like data volumes i also create rollback scripts for critical migrations, though i prefer rolling forward with fixes for complex migrations, i implement feature flags to control the transition all migrations are performed during low traffic periods with comprehensive monitoring and explicit approval processes after migration, i verify data integrity with automated checks " mediocre response "i use a migration tool like flyway or knex migrations to manage database changes each migration is versioned and applied in order before deploying to production, migrations are tested in staging environments for large tables, i'm careful about adding indexes or making changes that could lock the table for too long we schedule migrations during off peak hours to minimize impact on users we have backup procedures in place in case something goes wrong " poor response "we use scripts to apply database changes during deployment i make sure to test the migrations in our development environment first for important changes, we back up the database before running the migration if something goes wrong, we can restore from backup we usually schedule a maintenance window for significant database changes to avoid affecting users " 11\ how do you approach performance optimization for a backend service? great response "i take a methodical, data driven approach to performance optimization i start by establishing clear performance metrics and slos for the service i use profiling tools and distributed tracing to identify bottlenecks rather than relying on intuition for database optimization, i analyze query plans, add appropriate indexes, and optimize queries based on execution plans i implement caching strategically at multiple levels—object caching, query results, and http response caching with proper invalidation strategies for api performance, i implement pagination, field filtering, and compression i optimize the data access layer with connection pooling and batch operations where appropriate for cpu bound operations, i implement asynchronous processing and consider language specific optimizations i continuously monitor performance metrics in production and implement percentile based alerting for slo violations most importantly, i validate optimizations with benchmarks to ensure they actually improve performance in realistic scenarios and avoid premature optimization that might complicate the codebase without significant benefits " mediocre response "i start by identifying the slowest parts of the application using profiling tools or monitoring data for database performance, i look at slow queries and add indexes where needed i implement caching for frequently accessed data to reduce database load i make sure connections are properly pooled and resources are released when not needed i also look for n+1 query issues and optimize them with batch loading if there are cpu intensive operations, i consider moving them to background jobs " poor response "when performance is an issue, i look at the code that's running slowly and try to optimize it i add caching where possible and make sure database queries are efficient if a particular endpoint is slow, i might add more server resources to handle the load i also make sure we're not loading more data than necessary from the database for complex operations, we might need to upgrade our infrastructure or add more capacity " 12\ explain your approach to microservices architecture and when you would choose it over a monolithic architecture great response "i view microservices as a trade off between organizational scalability and technical complexity i'd choose microservices when we need independent scaling of components, have teams that need to work autonomously, or have distinct bounded contexts with clear boundaries before adopting microservices, i ensure we have the necessary infrastructure for deployment automation, service discovery, monitoring, and distributed tracing for implementation, i focus on clear service boundaries following domain driven design principles, avoid shared databases, and implement resilience patterns like circuit breakers and bulkheads i'm careful about breaking changes and implement contract testing between services i also consider the data consistency challenges in distributed systems, implementing patterns like saga for distributed transactions when necessary in contrast, i'd choose a monolith for simpler domains, smaller teams, or early stage products where development speed is more important than scalability the most effective approach is often starting with a modular monolith with well defined boundaries that can later be extracted into microservices if needed " mediocre response "i'd choose microservices when the application needs to scale different components independently or when multiple teams need to work on different parts of the system simultaneously microservices allow for more technology diversity and fault isolation however, they come with overhead in terms of deployment complexity and network communication for smaller applications or when starting a new project, i'd probably go with a monolith and extract microservices when specific needs arise the important thing is having clear module boundaries regardless of the architecture " poor response "microservices are the modern way to build scalable applications they're better than monoliths because each service can be deployed independently, which speeds up development i'd use microservices for any significant project because they're more maintainable in the long run with tools like docker and kubernetes, it's easier than ever to deploy and manage microservices each team can work on their own service without affecting others " 13\ how do you handle authentication and authorization in a backend system? great response "i implement authentication using industry standards like oauth 2 0 and openid connect, preferably with an established identity provider rather than rolling our own for authentication flow, i use authorization code flow with pkce for web applications and appropriate flows for other client types token validation happens at the api gateway level, with short lived access tokens and secure refresh token rotation for authorization, i implement a role based access control system enhanced with attribute based controls for fine grained permissions i store permissions in a distributed cache for fast access for sensitive operations, i implement additional verification steps all authentication and authorization decisions are thoroughly logged for audit purposes i also implement security headers, csrf protection, and rate limiting to prevent common attacks for microservices, i propagate authenticated user context via signed jwts or through a secure token exchange, and implement service to service authentication using mtls or api keys with appropriate rate limits " mediocre response "i implement authentication using jwt tokens with proper signature validation users authenticate with credentials or oauth providers, receiving a token that's sent with subsequent requests for authorization, i implement role based access control where each endpoint checks if the user has the required permissions i store user roles and permissions in the database and possibly cache them for performance for sensitive operations, we might require re authentication i make sure tokens expire appropriately and implement refresh token flows " poor response "i typically use jwt tokens for authentication since they're stateless and easy to implement users login with username and password, and we return a token they include in the authorization header for authorization, we check if the user has the right role to access each endpoint we store user roles in the database and check them on each request we set reasonable expiration times on tokens so users don't have to login too frequently " 14\ how do you handle data consistency in distributed systems? great response "i approach data consistency in distributed systems by first understanding the specific consistency requirements for each use case using the cap theorem as a framework for most scenarios, i implement eventual consistency with clear event sequencing using techniques like logical clocks or event versioning i use the outbox pattern with idempotent consumers to ensure reliable event propagation across services for data that requires stronger consistency, i implement distributed transactions using the saga pattern, choreographing compensating transactions for rollbacks i carefully design conflict resolution strategies for concurrent updates, often using crdts where applicable or last write wins with vector clocks to detect inconsistencies, i implement reconciliation processes that periodically verify data across systems i also use change data capture to propagate database changes reliably throughout the system, i maintain clear visibility into consistency states through monitoring and alerting on reconciliation metrics " mediocre response "i use eventual consistency for most distributed operations since it provides better availability when immediate consistency is required, i implement two phase commits or saga patterns for distributed transactions i make operations idempotent so they can be safely retried, and i use unique ids to detect duplicates for conflict resolution, i typically use timestamp based approaches or application specific conflict resolution it's important to design the system to handle temporary inconsistencies and provide ways to detect and resolve them " poor response "i try to use transactions whenever possible to maintain consistency when working across services, i make sure operations happen in the correct order and implement retry logic for failures if a service is down, we queue the requests and process them when it's available again for most cases, eventual consistency is good enough, and users usually don't notice small delays in data propagation if there are conflicts, we usually go with the most recent update " 15\ how would you design a system for handling high throughput event processing? great response "i'd design an event driven architecture using a robust message broker like kafka or aws kinesis as the backbone i'd partition events by key to enable parallel processing while maintaining order when necessary for the processing layer, i'd implement stateless consumers that can scale horizontally, with auto scaling based on queue depth metrics i'd design idempotent consumers to handle duplicate events safely, using deduplication techniques like idempotency keys or exactly once semantics where available for windowed operations or aggregations, i'd use a stream processing framework like kafka streams or flink to handle backpressure, i'd implement rate limiting and circuit breakers for reliability, i'd add dead letter queues with automated retries using exponential backoff i'd also implement comprehensive monitoring with metrics on processing rates, latency percentiles, error rates, and consumer lag, with alerts for processing delays for extremely high throughput, i'd consider approaches like batch processing of events and optimizing serialization formats using protocols like protocol buffers or avro " mediocre response "i'd use a message queue or streaming platform like kafka for high throughput events i'd design the system to process events asynchronously and scale horizontally by adding more consumer instances i'd make sure events are partitioned effectively to distribute the load the consumers would be designed to be idempotent to handle potential duplicate events i'd implement monitoring to track queue depths and processing times, with alerts for abnormal patterns " poor response "i'd use a queue system like rabbitmq or sqs to buffer the events and have worker processes consume them if we need more throughput, we can add more workers i'd make sure to have good error handling so failed events don't block the whole system we'd track metrics to make sure we're keeping up with the incoming events if the system gets overloaded, we might need to add more servers or optimize the processing code " 16\ how do you approach database indexing and query optimization? great response "i approach database indexing methodically, starting with understanding access patterns by analyzing query execution plans and slow query logs i create indexes based on where conditions, join clauses, and order by statements, prioritizing high impact queries identified through profiling for composite indexes, i consider column selectivity and arrange columns strategically to maximize their effectiveness for multiple query patterns i'm careful about over indexing, which can degrade write performance and increase storage requirements beyond indexing, i optimize queries by rewriting them to use exists instead of in for subqueries where appropriate, using explain analyze to compare approaches, and ensuring proper join types i implement query denormalization for read heavy scenarios, using materialized views or summary tables with appropriate refresh strategies for large tables, i implement partitioning strategies based on access patterns i also utilize database specific optimizations like partial indexes in postgresql or covering indexes in sql server finally, i set up continuous monitoring of query performance and index usage statistics to identify opportunities for further optimization " mediocre response "i analyze slow queries using the database's explain feature to understand execution plans i add indexes on columns used in where clauses, join conditions, and sorting operations for composite indexes, i try to order columns from highest to lowest selectivity i'm careful not to add too many indexes since they slow down writes and take up space besides indexing, i look for inefficient query patterns like n+1 queries or unnecessary joins and optimize them i also consider database specific features like partial indexes or include columns when appropriate " poor response "when queries are slow, i look at what fields are being used in where clauses and add indexes on those columns i use the explain command to see if the indexes are being used if tables are very large, i might add indexes on other columns used in joins or order by if performance is still an issue, i might denormalize some data to reduce the need for complex joins i try not to add too many indexes since they make inserts and updates slower " 17\ how do you monitor and troubleshoot performance issues in a production backend system? great response "i implement a comprehensive observability stack with metrics, logs, and distributed tracing for metrics, i track key performance indicators like response time percentiles (not just averages), error rates, and resource utilization across services i use distributed tracing with tools like jaeger or aws x ray to track requests across service boundaries, identifying bottlenecks in the request path i implement structured logging with correlation ids to connect logs across services for alerting, i focus on user impacting slos rather than just resource metrics, using appropriate thresholds based on historical patterns and percentiles to reduce false positives when troubleshooting, i follow a methodical approach first identifying whether the issue is systemic or isolated using dashboards, then narrowing down to specific services using distributed traces, and finally examining detailed logs or executing targeted profiling in production when necessary i also implement synthetic transactions and real user monitoring to proactively detect issues before they impact many users after resolving issues, i conduct thorough post mortems to identify root causes and implement changes to prevent recurrence " mediocre response "i set up monitoring tools like prometheus and grafana to track key metrics like response times, error rates, and resource utilization i implement logging with a centralized log management system so we can search and analyze logs efficiently when issues occur, i look at the dashboards to identify which components are affected, then dive into the logs for more details i set up alerts for important thresholds, like high error rates or elevated response times for distributed systems, i use distributed tracing to follow requests across different services " poor response "we have monitoring tools that track cpu, memory, and response times for our services when users report problems, i check the logs to see what's happening if something is running slowly, i look at the database queries that might be causing the issue or check if there's high server load we also have alerts set up for when error rates exceed normal levels for complex issues, we might need to add more logging temporarily to understand what's happening " 18\ how would you implement a robust database backup and recovery strategy? great response "i implement a multi layered backup strategy that balances recovery point objectives (rpo) and recovery time objectives (rto) with operational costs i use a combination of full backups at regular intervals (typically daily) and incremental or differential backups more frequently, complemented by continuous transaction log backups for point in time recovery capability all backups are automatically tested for integrity and recovery testing is performed regularly with automated restoration to verify recoverability i implement multi region replication of backup data with appropriate encryption at rest and in transit for critical systems, i maintain standby databases using synchronous or asynchronous replication depending on performance requirements i document and regularly practice the recovery procedures, including partial restoration scenarios i also implement monitoring for backup jobs with alerts for failures or missed backups for large databases, i use techniques like snapshot based backups to minimize impact on production performance the entire strategy is driven by clearly defined rpo/rto targets that are regularly reviewed with stakeholders to ensure they align with business needs " mediocre response "i implement daily full backups and more frequent transaction log backups to enable point in time recovery backups are stored both locally and in a remote location or cloud storage for redundancy i regularly test restores to verify backup integrity for critical systems, i set up database replication to a standby server that can take over if the primary fails i document recovery procedures so they can be followed even under pressure i make sure backups are monitored and alerts are set up for failures " poor response "i set up automated daily backups of the database and store them for at least a month we keep the backups in a different location from the production database before major changes, we take additional backups just in case if there's a failure, we can restore from the most recent backup for important databases, we might also set up replication so we have a standby copy if needed " 19\ how do you approach writing maintainable and scalable code? great response "i prioritize code maintainability through several practices i use solid principles to create modular, cohesive components with clear responsibilities i implement a clean architecture that separates business logic from infrastructure concerns, making it easier to modify either independently i'm deliberate about dependency management, using dependency injection and designing for testability for scalability, i identify potential bottlenecks early and design with horizontal scaling in mind—using stateless components where possible and implementing appropriate caching strategies i write comprehensive automated tests at multiple levels to enable confident refactoring i'm also a strong believer in consistent code style and documentation of non obvious decisions using architectural decision records (adrs) to preserve context for future developers i practice continuous refactoring to prevent technical debt accumulation and regularly review the codebase for improvement opportunities i also consider operational aspects during development, implementing health checks, metrics, and graceful degradation to ensure the system remains resilient under load " mediocre response "i follow solid principles and design patterns to keep code modular and maintainable i write clean, self documenting code with meaningful variable and function names i make sure to include unit tests for the key functionality to catch regressions i try to avoid premature optimization but design with scalability in mind, keeping components stateless when possible and identifying potential bottlenecks early i review code regularly and refactor when necessary to prevent technical debt from accumulating " poor response "i focus on getting the code working first, then clean it up if there's time i try to reuse existing code wherever possible to avoid duplicating effort i add comments to explain complex parts of the code when i notice something that could be improved, i make a note to come back to it later for scalability, i make sure the database is properly indexed and add caching for frequently accessed data "