CS Undergrad Builds Production-Grade Distributed API Gateway From Scratch

·1 views

A computer science undergraduate has developed a fully functional, production-grade API gateway built on Node.js, featuring reverse proxy routing to multiple microservices and per-client rate limiting using a Redis token-bucket algorithm. The system enforces JWT authentication at the gateway level and handles asynchronous request logging through BullMQ worker queues connected to a MongoDB database. Live traffic metrics are streamed to a React dashboard via Socket.io, displaying data such as total requests, requests per minute, average response time, and error rate. The project was built in seven phases, progressing from basic reverse proxy setup through authentication, rate limiting, async analytics, and a real-time frontend. The developer has published the project on GitHub and is seeking community feedback on the queue worker design and rate limiting strategy.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

How Developers Can Efficiently Transcribe Audio Files Into Searchable Text

Developers often need audio recordings converted to text because text is searchable, shareable, and can be fed into tools like LLMs or subtitle generators. Three main transcription approaches exist: fully automated models such as OpenAI Whisper, manual typing for high-accuracy needs, or a hybrid method combining automation with a light editorial review. Audio quality preparation is critical — converting files to mono WAV at 16 kHz using tools like ffmpeg significantly reduces transcription errors and hallucinations. Whisper can be run locally for privacy and cost savings, or accessed via OpenAI's cloud API for faster, zero-setup processing. For technical content containing proper nouns or CLI commands, using larger Whisper models improves accuracy, and leveraging timestamped JSON output enables subtitle generation and precise audio referencing.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Developer builds free interactive map of 767 global train routes using open data

A solo developer has launched TrainRouter, a free, no-signup interactive map cataloguing 767 notable train routes across 118 countries, including iconic services like the Trans-Siberian Railway and the Glacier Express. Built using React and MapLibre GL JS on OpenFreeMap vector tiles, the project required no API keys and incurs no per-tile hosting costs. Each route carries detailed metadata including distance, journey time, top speed, operator, and rolling stock, all derived from a single flat data file. To ensure search engine and AI crawler visibility, the developer pre-rendered around 1,500 static HTML pages at build time, since single-page apps are otherwise invisible to most crawlers. The project also exposes its dataset as open data, enabling queries such as ranking countries by their fastest train's average end-to-end speed rather than headline top speed.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Developer's Practical Guide to Transcribing Audio Using Whisper and Hybrid Workflows

Developers working with meeting recordings, podcasts, or interviews can convert audio to searchable, structured text using a combination of automated and manual transcription methods. Tools like OpenAI's Whisper model or cloud-based speech-to-text APIs can handle most transcription work quickly, though accuracy drops with background noise, crosstalk, or heavy accents. A recommended hybrid approach involves running automated transcription first to cover 80–90% of the work, then manually editing the output for accuracy. Pre-processing audio with tools like ffmpeg — standardizing to mono, 16 kHz WAV or FLAC — significantly improves model performance. Once transcribed, text outputs can be used for documentation, RAG pipelines, Elasticsearch indexing, accessibility improvements, or feeding into LLM context windows.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Developer Builds Lightweight Open-Source Anti-Stealer Tool for Windows in Python

A developer frustrated with resource-heavy commercial security tools has created a standalone behavioral anti-stealer application aimed at blocking modern infostealers such as Lumma and RedLine. The tool is written in pure Python and compiles to a roughly 9MB executable, consuming only 0–2% CPU during testing, with no cloud telemetry or signature databases. It operates by scanning active network connections every five seconds, using process lineage validation, SHA-256 integrity checks, and file descriptor auditing to detect suspicious access to browser and messaging app data. The project is released under the GNU GPL v3 license and is publicly available on GitHub for community review. The developer is seeking feedback on whether the five-second detection interval is sufficient against fast-executing infostealers and on potential false-positive risks in Windows process tracking.

0 comments Read more at DEV Community