Where developers grow together

KernelPulse

KernelPulse reimagines Linux server monitoring for people who don't want to run a monitoring stack just to monitor a server. Its purpose is to strip away the operational weight of Prometheus, exporters, Grafana, and a separate database, and replace all of it with one Python process that reads the kernel directly and ships with real intelligence built in — not bolted on. The goal isn't just visibility, it's judgment. Most lightweight dashboards show you numbers and leave the interpretation to you; most tools that do interpret the numbers require standing up real infrastructure to get there. KernelPulse closes that gap with a transparent AI layer — a z-score anomaly classifier, a slow memory-leak detector that catches growth patterns z-score alone would miss, and a capacity-exhaustion forecaster that tells you "RAM full in ~2h 14m" instead of just "RAM: 87%." Every one of those calls is explainable: the dashboard shows the raw value, the moving average, and the exact statistical threshold behind each status, so nothing is a black box. Ultimately, KernelPulse is built for the developer, student, or small team running a personal server or home lab who wants Prometheus-grade insight without Prometheus-grade overhead — a monitoring tool that installs in one command and actually tells you when something needs your attention, not just what the numbers are.

RecallOS

RecallOS is an AI-powered memory layer for your browser. It passively captures every page you visit, uses AI to summarize, tag, and semantically embed it, and lets you retrieve it later with natural language and not exact filenames or keywords. The goal is simple: turn fragmented, forgettable browsing into a searchable, context-aware knowledge layer, so that finding something you've previously seen is as easy as describing it in your own words without manual saving, tagging, or organizing required.

ZENITH

Zenith is an AI-powered startup validation tool focused on the legal-tech vertical. Founders enter their startup idea, and the system generates a structured Ideal Customer Profile covering segment, budget sensitivity, and key objections. It then simulates conversations with four distinct synthetic personas — solo advocate, firm partner, in-house counsel, and client — each responding dynamically through an LLM with real pushback, not scripted agreement. Founders can chat directly with these personas in a fully functional adversarial interview format to stress-test their idea. An insights panel then summarizes recurring pain points and feature suggestions from the conversation, clearly labeled "synthetic-only" to signal it's simulated feedback rather than real user data. Built with FastAPI, LangGraph, and Next.js, Zenith turns early-stage idea validation into a fast, structured, and repeatable process.