Serhii Soldatov’s Post

API was slow. Everyone blamed .NET.” Turns out, .NET wasn’t the villain. The database was. A single missing index was turning a 300ms query into a 4-second bottleneck. One line fixed it all: CREATE INDEX IX_UserId ON Orders(UserId); Lesson learned? Before you optimize code — optimize what your code talks to. Backend issues often live one layer below where we’re looking.

If there's one thing I've learned about performance issues over 25 years, it is that the performance bottleneck is always somewhere else than I assume. 🤭👌 Very often go for bugs too.

Node.js Streams Explained What Moving the Pyramids Can Teach You About Node.js Streams 🏛️ Imagine you have a 3GB file that you need to upload/download using your Node.js server. How do you solve this task? Your first instinct as a Node.js developer might be to use an npm package. Sounds great to me! But what if I told you that while you're treating the file, you also need to add some business logic? Let's say, for example, the file is a CSV, and you need to keep certain lines and ignore others—while saving the ignored lines in a separate file. Can your npm package handle that? Probably. Is it easier, simpler, scalable? Probably yes. But now you need another package and more complex code in your codebase, which might affect performance. Now let's take a step back. Before downloading the package, did you ask yourself what technique the creators of these packages use under the hood? I'll answer that for you—it's most likely streams, and that's what we're going to talk about in this post. So buckle up, and let https://lnkd.in/gRjYzC_z

Node.js Streams Explained What Moving the Pyramids Can Teach You About Node.js Streams 🏛️ Imagine you have a 3GB file that you need to upload/download using your Node.js server. How do you solve this task? Your first instinct as a Node.js developer might be to use an npm package. Sounds great to me! But what if I told you that while you're treating the file, you also need to add some business logic? Let's say, for example, the file is a CSV, and you need to keep certain lines and ignore others—while saving the ignored lines in a separate file. Can your npm package handle that? Probably. Is it easier, simpler, scalable? Probably yes. But now you need another package and more complex code in your codebase, which might affect performance. Now let's take a step back. Before downloading the package, did you ask yourself what technique the creators of these packages use under the hood? I'll answer that for you—it's most likely streams, and that's what we're going to talk about in this post. So buckle up, and let https://lnkd.in/gRjYzC_z

Wanted to highlight some Jupiterp (www.jupiterp.com) improvements I've been working on for the past couple months and I don't know where else to write it: - Migrated course, section, and instructor data to SQL and created a public API (api.jupiterp.com) - Created an open-source TypeScript SDK (npm install @jupiterp/jupiterp) for using the new API and integrated it into Jupiterp client-side - Implemented client-side caching to largely eliminate loads for already-fetched data - Implemented server-side caching, reducing time to respond on server from ~100 to ~5 ms for cached responses - Added filters to course searching - you can now filter your searches by GenEd, instructor, credit amount, and seat availability. This was really the ultimate goal of all of the previous changes: improved search experience with these filters! More features to come!

Meet js-kt (server) + kt-client (type-safe client) — a one-two punch that turns API contracts from fragile guesses into first-class, editor-backed truth. Why this matters: ✅ No more “what does the backend expect?” — Live types from server → client, always. ⚡ Blazing routing + clustering — O(1) lookups, pattern caching, and multi-core support out of the box. 🔁 One handler, two transports — serve the same logic over HTTP and WebSockets with zero duplication. 🧭 File-based routes = zero boilerplate — your folder layout is the API. 📦 Auto-docs & type generation — npx kt-client load-types gives you editor autocomplete and runtime safety. Why I built it: I wanted a backend stack where developer experience and production performance aren’t compromises — clean DX, strict types, and rock-solid runtime behavior. If you build APIs, real-time features, or SPAs that rely on tight contracts — this will save you debugging hours and shipping risk. 🔗 Try it / star it / break it: https://lnkd.in/dbTjHebd https://lnkd.in/d-FPta5K #TypeScript #NodeJS #WebSockets #APIDesign #OpenSource #DeveloperExperience #jskt #ktclient

💡Why HTTP Status Codes Matter More Than You Think Ever seen a response like 404 or 500 and just moved on? Well, these numbers quietly define how your backend talks to the frontend. Here’s a quick breakdown 👇 ✅ 200 – OK → Everything worked fine ✅ 201 – Created → New resource successfully created ⚠️ 400 – Bad Request → Client sent invalid data 🚫 401 – Unauthorized → Authentication required 🚫 403 – Forbidden → You’re authenticated, but not allowed ❌ 404 – Not Found → The resource doesn’t exist 💥 500 – Internal Server Error → Something broke on the server A well-structured backend isn’t just about data — it’s about communication clarity. Returning the right status code makes debugging easier and APIs more reliable. 👉 Always respond with meaningful status codes and messages. #MERN #WebDevelopment #Backend #MERNStack #HTTP #API #CleanCode

🚀 Sometimes limited resources push you to build smarter, not bigger. In one of my recent projects, I had to make a legacy Windows application send data to a Laravel backend — in real time. The Windows app itself wasn’t built by me, but I developed a separate Go service that collected its data, structured it in JSON, and sent it to the backend continuously. Our biggest challenge? Limited server capacity and a small team — but real-time performance was a must. So instead of sending one request per account, I batched all account data into a single request and moved part of the computation to the Windows side. To avoid losing data if a request failed, I handled the heavy processing and retry logic inside the Go service. Each batch included a status flag, so the backend knew what stage the data was in — keeping everything consistent and reliable even under failure. The result: ✅ Fewer API calls ✅ Lower load on the Laravel backend ✅ Stable and fault-tolerant real-time performance This project reminded me that you don’t always need a perfect setup — sometimes, working smart with what you have leads to the best solutions. 💡 As we scale, the next step is to evolve this setup into a more distributed and scalable architecture. Have you ever had to design around tight limitations? I’d love to hear your story 🙌

🚀 My Second Project with Docker – Django + MySQL + NGINX Load Balancer In this setup, I’ve containerized a Django-based application and built a small scalable architecture on my local machine using Docker. Here’s how it’s structured: 🐳 Containers Setup: 1 MySQL container running on port 3306 3 Django containers (from image wliari/animelist_image) running on ports 8000, 8001, and 8002 All containers are connected using a custom Docker network 🌐 NGINX Reverse Proxy / Load Balancer: NGINX listens on port 80 Requests to / are load-balanced (round-robin) across the three Django containers This ensures high availability and even traffic distribution 🧩 Key Takeaway: This project helped me understand how Docker networking, reverse proxying, and load balancing can work together seamlessly — even on a local environment. Here’s the architecture overview 👇

# Self-Hosted Push Notifications Part-2 Self-Hosted Push Notifications Specification Part 2: Backend Implementation \(Go\) Version: 1.0 Last Updated: October 2025 Prerequisites: Part 1: Architecture & Setup Author: Bunty9 License: MIT \(Free to use and adapt\) Dependencies GORM Models Type Definitions Push Service Implementation HTTP Controllers Route Configuration Middleware Utilities Main Application Testing # Navigate to backend directory cd backend # Initialize Go module \(if not already done\) go mod init https://lnkd.in/gTYwJhFK # Install core dependencies go get github.com/gin-gonic/gin # HTTP framework go get gorm.io/gorm # ORM go get gorm.io/driver/postgres # PostgreSQL driver go get https://lnkd.in/g5HHa3i6 # Web Push Protocol go get github.com/google/uuid # UUID generation go get github.com/joho/godotenv # Environment variables go get github.com/golang-jwt/jw https://lnkd.in/gKTpB9p6