Magomed Bonkurov

Java 8–17

The language I have written in since 2019.

How I use it I went from Java 8 to 17 on real projects, including an 11 → 17 migration in a product team. Of the newer features I use records for DTOs, text blocks for SQL and switch expressions most often.

Java 21 (familiar)

The latest LTS release, with virtual threads.

How I use it I have not shipped it to production, and I say so plainly. I have read about virtual threads and pattern matching and tried them on my own tasks, but I have no experience running them under load.

SQL

The query language for relational databases.

How I use it I write it by hand rather than trusting whatever the ORM generates. I read execution plans regularly: almost every performance problem I have fixed turned out to be in the query, not in the Java code.

OOP, SOLID

The principles for splitting code into parts.

How I use it I treat them as a tool, not a doctrine. In practice two of them pay off most: one reason to change per class, and depending on an interface rather than an implementation. Those are the cheapest to maintain.

Concurrency

Running several things at the same time.

How I use it I have debugged race conditions and deadlocks in production. I know the price: more often than not the right answer is to remove shared state rather than add threads. I have worked with thread pools, synchronisation and atomics.

Algorithms & data structures

The ground covered in technical interviews.

How I use it I practise regularly to stay in shape for interviews. At work the useful part is rarely exotic — it is understanding why a query inside a loop means O(n) round trips to the database, and how to collapse it into one.

Spring Boot

The framework almost every Java service is built on.

How I use it My main tool for all six years. I have started services from scratch and joined existing ones; I understand how auto-configuration works and can explain where a bean I never declared came from.

Spring Data JPA

A data access layer: objects instead of hand-written SQL.

How I use it I use it where it saves time. I also know its weak spot — a silent N+1 and pulling entire object graphs into memory. That is exactly what I fought when I made the assignment endpoint faster.

Spring Security

Authentication and access control.

How I use it I set up JWT login and a student / teacher / administrator role model, including blocking access to other people's data — the most common hole in systems like that.

Spring MVC

The web controller layer in Spring.

How I use it I write REST endpoints, error handling and input validation.

Hibernate

An ORM: maps database tables to objects in code.

How I use it I work with it and I know when to step away from it: if I need three fields and it loads an entire entity graph, I write the query directly.

JdbcTemplate

Direct SQL without an ORM.

How I use it My tool for heavy queries. I moved several reporting queries onto it where mapping into entities was wasted work, and report generation got several times faster.

Resilience4j

Protection against failures in neighbouring services.

How I use it I wrap outgoing synchronous calls in it: timeout and circuit breaker. The point is that another team's outage should not eat my thread pool and take my endpoint down with it.

PostgreSQL

The database on every project I have worked on.

How I use it I do not only write queries — I work on how they execute: indexes, partitioning, locks, isolation levels. Most of my performance work happened here rather than in Java.

EXPLAIN ANALYZE

Shows how the database actually runs a query.

How I use it The first thing I open when something is slow. I look at the pages read as well as the time: often the index exists but the planner ignores it, and this is the only place that shows it.

pg_stat_statements

Statistics across every query hitting the database.

How I use it I use it to find the real load rather than the load everyone assumes. More than once the slow thing was not the query people complained about, but a quiet neighbour running a hundred times more often.

Indexing, partitioning

Two ways to avoid reading data you do not need.

How I use it I design composite indexes around the actual query predicates and verify the planner uses them. I applied partitioning to an events table that had grown to hundreds of millions of rows, moving the history in batches with no downtime.

Materialized views

A query result computed ahead of time and stored.

How I use it I move heavy analytics into them and refresh on a schedule. The trade-off is clear: the data is no longer fresh to the second, but reports stop competing with live traffic for the database.

Redis

A fast in-memory store, usually used for caching.

How I use it I cache profiles and reference data using cache-aside. I keep a TTL even with event-based invalidation, as insurance in case an event is lost, and I spread the TTLs so keys do not all expire at once during a peak.

MySQL

Another relational database.

How I use it I worked with it on earlier projects. My real depth is in PostgreSQL.

Flyway / Liquibase

Version control for the database schema.

How I use it Every schema change ships as a migration alongside the code. The rule is simple: nobody edits the production database by hand, otherwise environments drift apart and it surfaces at the worst possible moment.

REST / OpenAPI

The main way services and clients talk to each other.

How I use it I design the resources, status codes and error format, and describe the contract in OpenAPI. Other teams generate their clients from that description instead of coming to me with questions.

Apache Kafka

A broker: services exchange events without waiting for each other.

How I use it I moved synchronous integrations onto it so that a neighbouring outage would not break my service. I keep the key fact in mind: delivery is at-least-once, so duplicates are unavoidable and the consumer must be idempotent.

GraphQL

The client asks for exactly the fields it needs.

How I use it I built the layer for a web and a mobile client with different data needs. Two things are mandatory: batching through DataLoader against N+1 in resolvers, and a limit on query complexity, or one client can take the service down.

Keycloak, OAuth2 / OIDC

An external service for login and tokens.

How I use it I moved authorisation onto it with a role model, so that no service has to store passwords or invent its own permission checks.

Transactional outbox

A pattern that stops events being lost on failure.

How I use it The event is written to a helper table in the same transaction as the data, so atomicity comes from the database rather than from application code. A separate process publishes it to the broker afterwards. This covers the case where the service dies between the commit and the publish.

Idempotency, DLQ

Reprocessing does not corrupt data, and failures do not loop forever.

How I use it I keep a table of processed events and drop duplicates by key. Anything that still fails after several attempts goes to a separate queue for manual review — otherwise one bad message blocks the whole partition.

JUnit 5

The standard testing tool in Java.

How I use it I cover business logic and edge cases. I am relaxed about coverage percentages: what matters is that the tests fail when something is genuinely broken, not that the number looks good.

Mockito

Replaces dependencies in tests.

How I use it I use it to isolate the class under test. I try not to mock anything I could run for real instead.

Testcontainers

Starts a real database and broker in Docker for the tests.

How I use it Far more reliable than mocks: the test runs against a real PostgreSQL, so it also checks the SQL, the migrations and the transaction behaviour. That catches a whole class of bugs mocks never see.

JMeter

Generates load to find where a service breaks.

How I use it I run ramp-up scenarios and watch what gives way first. In my experience it was usually not the Java code but the database connection pool size or a synchronous outbound call.

Integration tests

Test the service as a whole rather than one class.

How I use it I find them more valuable than unit tests for the part of the code where things actually break: database queries, transactions, and contracts between services.

Docker

Packages a service together with its environment.

How I use it I build images and start a local database and broker through compose, so my machine and the test environment run the same thing.

GitLab CI/CD

Automated build, tests and deployment.

How I use it I set up pipelines with a quality gate: the build fails on critical analyser findings and when coverage drops. The point is that quality is checked by a machine rather than by whoever on the team is paying most attention.

Kubernetes (operating)

An orchestrator: runs and keeps services alive in a cluster.

How I use it To be precise: I operate my services in it — deployments, config maps, secrets, and diagnosing problems through logs and describe. The cluster itself is run by a platform team.

Grafana

Dashboards for load, latency and errors.

How I use it I come here after every optimisation: a number from a load test is worth nothing until the effect shows up in production at peak hour.

Kibana / ELK

Search across the logs of every service at once.

How I use it My main tool during an incident: I reconstruct what happened from the logs and then reproduce the problem in a test environment.

Nginx

A web server and reverse proxy.

How I use it I have configured static file serving, proxying to services and HTTPS.

Linux

The system all of this runs on.

How I use it Comfortable in the terminal: logs, processes, resource usage and networking, whenever I need to understand what is happening on a server.

Git, GitLab, Bitbucket

Version control and working with other people.

How I use it Branches, reviews, resolving conflicts, tracking a regression back through history. I review code and explain my comments rather than just rejecting the change.