Engineering With Java: Digest #84
LLM with Java, CQRS, Bloom filters, Cron jobs, JVM profiler and much more
👋 Java Devs! Welcome to this week’s addition! I hope you’re all doing great.
📢 I started another publication that focuses on SQL and Database & covers Daily SQL question. If it interests you Consider subscribing
This week, we cover essential insights on:
📢 Get actionable Java and Spring Boot insights every week, including practical code tips and real-world, use-case-based interview questions, to help you level up your backend skills—join 6,700+ subscribers for hand-crafted, no-fluff content.
First 100 paid subscribers will get the annual membership at $50/year forever that is ~ $4/mo ( 76 already converted to paid, 24 remaining)
Testimonials
🗒️ Articles Of The Week (8)
Training a Neural Network Model With Java and TensorFlow
This article shows how to train a neural network in Java using TensorFlow’s Java API, covering model setup, data preparation, and training flow. It explains core concepts like forward and backpropagation and demonstrates Java’s role mainly for running and integrating ML models in production rather than research.
Optimizing Java Back-End Performance Profiling and Best Practices
This article emphasizes starting Java performance optimization with profiling tools like JFR and JMC to identify real bottlenecks such as GC pressure, N+1 queries, and thread contention. It then outlines fixes like reducing object creation, optimizing database queries, and tuning JVM and concurrency settings. The key takeaway is that performance tuning must be holistic across code, database, and JVM rather than isolated changes.
Bloom Filters for Blocking Repeated Cache Misses in Spring Boot
This article explains how Bloom filters can be used in Spring Boot to prevent repeated cache misses by acting as a fast pre-check before querying Redis or the database. It stores hashed key positions in a compact structure to quickly reject non-existent data and avoid repeated expensive lookups. This improves performance in high-traffic systems by reducing unnecessary cache/database calls at the cost of a small false-positive rate.
CQRS in Java: Separating Reads and Writes Cleanly
This article explains CQRS in Java, where read and write operations are split into separate models to improve clarity, scalability, and maintainability. It shows how Spring Boot can implement this using separate services or repositories, with writes handling business logic and reads optimized for fast queries. It also notes the trade-off of added complexity and potential eventual consistency between the two sides.
How I Automated Weekly Twitter/X Posts With Java, JBang and GitHub Actions
This article describes automating weekly Twitter/X posts using Java, JBang, and GitHub Actions with no servers or external infrastructure. It uses YAML-based tweet storage, a queued/shuffled posting system, and Git-tracked state, while a JBang script posts via the Twitter API and updates status after success. The setup runs on a scheduled GitHub Actions workflow, making it fully automated, resumable, and low-cost.
Getting a Cron Expression From Database for a Spring Boot Scheduled Job
This article explains how to schedule Spring Boot tasks using cron expressions stored in a database instead of hardcoding them. It compares a static approach using SpEL with a Spring bean and a dynamic approach using SchedulingConfigurer that fetches the cron before each run. The key takeaway is that dynamic scheduling enables runtime updates without restarting the application, at the cost of added complexity.
The End of Destructive AI Hallucinations: Hybrid Kernel Architecture with Java 25 and Zero-Trust Guardrails
This article proposes a hybrid kernel architecture that separates deterministic operations from LLM-based tasks to reduce latency and hallucinations. It routes simple commands through traditional logic while using LLMs only for complex reasoning, protected by a zero-trust safety layer with safeguards like kill-switches and ephemeral branches. The approach claims significant performance gains and safer AI execution by preventing destructive or irreversible changes.
I Built a JVM Profiler That Points AI at Your Exact Line of Broken Code — Here’s How It Works Under the Hood
This article describes a JVM profiler that identifies performance or runtime issues at the exact line of code and uses an LLM to explain root causes and suggest fixes. It combines deterministic JVM analysis (heap, GC, threads) with structured AI input to reduce hallucinations and ensure safe, reliable insights. The goal is faster debugging by directly linking system metrics to problematic code lines in minutes.
▶️ Videos of the week (3)
How Netflix Uses Java - 2026 Edition
The talk explains that Netflix uses Java and JVM-based tech (mainly Spring Boot and Kotlin) across most backend systems, including streaming microservices, GraphQL (DGS), Open Connect, and data pipelines, due to its strong balance of performance, scalability, and maintainability. Other languages like Python, Go, and JavaScript are used only for specific needs such as ML, infrastructure, and UIs. It also highlights Netflix’s standardized Spring Boot platform and large-scale automation strategies for testing and upgrading thousands of services efficiently.
How JDK 26 Improves G1's Throughput
This episode explains how the G1 garbage collector works in Java, focusing on write barriers, card tables, and concurrent refinement used to track object references efficiently. It highlights the trade-off between latency and throughput and how G1 became the default GC. It also covers Java 26 improvements that reduce synchronization overhead, improving throughput and moving G1 closer to being the universal default collector.
Spring Boot in the Cloud: Advanced Optimization Deep Dive
This talk focuses on improving Spring Boot performance across build time, startup time, memory usage, image size, and throughput using the PetClinic app as a benchmark. It shows how JVM versions, Buildpacks/Paketo, and runtime choices impact results. The key takeaway is that performance optimization must be measurement-driven with proper benchmarking to avoid misleading results and safely reflect production behavior.
🔥 Recently Published In-house Blogs (4)
Spring Data Interview Question - Orders API Suddenly Slows Down
Java Interview Question - Detecting Duplicate Product Titles
Java Bug Fix Interview Question - Retry Utility with Silent Failures
Spring Boot Interview Question - Scaling User Search Endpoint
Thats all for this week friends! Thanks for reading this far. If you liked it please share with your network.
Happy Coding 🚀
Suraj
Subscribe | Sponsor us | LinkedIn | Twitter