<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Work on Muhammad Nuh</title><link>https://hexrogue.github.io/projects/</link><description>Recent content in Work on Muhammad Nuh</description><generator>Hugo</generator><language>en-us</language><managingEditor>ikanlemas0@gmail.com (Muhammad Nuh)</managingEditor><webMaster>ikanlemas0@gmail.com (Muhammad Nuh)</webMaster><lastBuildDate>Tue, 27 Oct 2026 00:00:00 +0800</lastBuildDate><atom:link href="https://hexrogue.github.io/projects/index.xml" rel="self" type="application/rss+xml"/><item><title>APEL.A Admissions Portal</title><link>https://hexrogue.github.io/projects/apela/</link><pubDate>Tue, 27 Oct 2026 00:00:00 +0800</pubDate><author>ikanlemas0@gmail.com (Muhammad Nuh)</author><guid>https://hexrogue.github.io/projects/apela/</guid><description>&lt;p&gt;Private client work. The source and candidate data are not public. This is the outline I can share.&lt;/p&gt;&#10;&lt;p&gt;APEL.A is Malaysia&amp;rsquo;s prior-learning assessment route for experienced professionals. The brief was to build the first APEL.A admissions system in the country: one portal where a candidate can build a portfolio, upload evidence, sit an assessment, book an interview, pay, and track the decision without chasing a folder of PDFs.&lt;/p&gt;&#10;&lt;h2 id="the-problem"&gt;The Problem&lt;/h2&gt;&#10;&lt;p&gt;The process was split across forms, email, spreadsheets, and staff memory. Candidates had to assemble a long portfolio from an official form. Assessors had to schedule exams and interviews across multiple calendars. Staff had to keep documents, payments, and status changes in sync. Every handoff was a place for something to go missing.&lt;/p&gt;</description></item><item><title>Admissions CRM</title><link>https://hexrogue.github.io/projects/crm/</link><pubDate>Sun, 20 Sep 2026 00:00:00 +0800</pubDate><author>ikanlemas0@gmail.com (Muhammad Nuh)</author><guid>https://hexrogue.github.io/projects/crm/</guid><description>&lt;p&gt;Private client work. Source and client data are not public. This is the outline I can share.&lt;/p&gt;&#10;&lt;h2 id="the-problem"&gt;The Problem&lt;/h2&gt;&#10;&lt;p&gt;The admissions team was running a postgraduate pipeline across spreadsheets, email threads, paper forms, and a folder of document templates. Every study plan, cover letter, and portfolio document was assembled by hand. The work was repetitive, easy to get wrong, and hard to hand over when someone was away.&lt;/p&gt;&#10;&lt;h2 id="what-i-built"&gt;What I Built&lt;/h2&gt;&#10;&lt;p&gt;A single admissions platform that owns the workflow from lead to document:&lt;/p&gt;</description></item><item><title>WhatsAPI</title><link>https://hexrogue.github.io/projects/whatsapi/</link><pubDate>Sun, 20 Sep 2026 00:00:00 +0800</pubDate><author>ikanlemas0@gmail.com (Muhammad Nuh)</author><guid>https://hexrogue.github.io/projects/whatsapi/</guid><description>&lt;p&gt;Private project. Source is not public. This is the outline I can share.&lt;/p&gt;&#10;&lt;h2 id="the-problem"&gt;The Problem&lt;/h2&gt;&#10;&lt;p&gt;The CRM needed WhatsApp messaging and calling without paying a third-party SaaS for every seat, every message, and every feature that should have been a webhook. The automation also needed to live close to the CRM, not in a vendor&amp;rsquo;s dashboard with its own ideas about data ownership.&lt;/p&gt;&#10;&lt;h2 id="what-i-built"&gt;What I Built&lt;/h2&gt;&#10;&lt;p&gt;A Go service that behaves like a proper gateway:&lt;/p&gt;</description></item><item><title>CareerGap</title><link>https://hexrogue.github.io/projects/careergap/</link><pubDate>Sun, 20 Sep 2026 00:00:00 +0800</pubDate><author>ikanlemas0@gmail.com (Muhammad Nuh)</author><guid>https://hexrogue.github.io/projects/careergap/</guid><description>&lt;p&gt;Private project. Source is not public. This is the outline I can share.&lt;/p&gt;&#10;&lt;h2 id="the-problem"&gt;The Problem&lt;/h2&gt;&#10;&lt;p&gt;Most skill-gap tools are either a keyword match with a nicer font or a language model asked to guess. Neither gives a candidate a clear, reproducible answer to a simple question: which requirements for this occupation are actually missing?&lt;/p&gt;&#10;&lt;h2 id="what-i-built"&gt;What I Built&lt;/h2&gt;&#10;&lt;p&gt;A deterministic analyzer that compares a candidate&amp;rsquo;s skills against occupation requirements from ESCO:&lt;/p&gt;</description></item><item><title>SKC Central</title><link>https://hexrogue.github.io/projects/skc-central/</link><pubDate>Tue, 27 Oct 2026 00:00:00 +0800</pubDate><author>ikanlemas0@gmail.com (Muhammad Nuh)</author><guid>https://hexrogue.github.io/projects/skc-central/</guid><description>&lt;p&gt;Private client work. Source and business data are not public. This is the outline I can share.&lt;/p&gt;&#10;&lt;p&gt;SKC Central is the internal operations platform behind a private education group. It brings finance, scheduling, teaching records, and admin into one Next.js application, so staff stop moving the same student data between spreadsheets and browser tabs.&lt;/p&gt;&#10;&lt;h2 id="the-problem"&gt;The Problem&lt;/h2&gt;&#10;&lt;p&gt;The group runs programmes, lecturers, agents, payments, commissions, timetables, and teaching files across several teams. The data lived in spreadsheets, exported reports, and a scheduling app that only covered one slice of the problem. Every team had its own version of the truth.&lt;/p&gt;</description></item><item><title>AdvisorNext</title><link>https://hexrogue.github.io/projects/advisornext/</link><pubDate>Sun, 20 Sep 2026 00:00:00 +0800</pubDate><author>ikanlemas0@gmail.com (Muhammad Nuh)</author><guid>https://hexrogue.github.io/projects/advisornext/</guid><description>&lt;p&gt;Private project. Source is not public. This is the outline I can share.&lt;/p&gt;&#10;&lt;h2 id="the-problem"&gt;The Problem&lt;/h2&gt;&#10;&lt;p&gt;Technical analysis across a watchlist is repetitive and easy to do inconsistently. One person looks at momentum, another looks at trend, a third ignores both and goes with a feeling. The goal was not to predict the market. The goal was to make the same inputs produce the same verdicts.&lt;/p&gt;&#10;&lt;h2 id="what-i-built"&gt;What I Built&lt;/h2&gt;&#10;&lt;p&gt;A dashboard that scores a watchlist against a fixed set of technical indicators:&lt;/p&gt;</description></item><item><title>SKC Advisor Chatbot</title><link>https://hexrogue.github.io/projects/skc-advisor-chatbot/</link><pubDate>Tue, 27 Oct 2026 00:00:00 +0800</pubDate><author>ikanlemas0@gmail.com (Muhammad Nuh)</author><guid>https://hexrogue.github.io/projects/skc-advisor-chatbot/</guid><description>&lt;p&gt;Private project. Source is not public. This is the outline I can share.&lt;/p&gt;&#10;&lt;p&gt;A WhatsApp-first admissions assistant for UNIRAZAK MBA and PhD programmes. It answers routine questions, keeps context across messages, and hands the conversation to a human advisor when the lead is ready.&lt;/p&gt;&#10;&lt;h2 id="what-i-built"&gt;What I Built&lt;/h2&gt;&#10;&lt;ul&gt;&#10;&lt;li&gt;A Cloudflare Worker with a pipeline: inbound webhook, AI response, intent detection, outbound reply.&lt;/li&gt;&#10;&lt;li&gt;A WhatsApp adapter layer so the provider can be swapped without rewriting the pipeline.&lt;/li&gt;&#10;&lt;li&gt;Multi-provider AI support behind one interface.&lt;/li&gt;&#10;&lt;li&gt;Per-user memory in Cloudflare KV.&lt;/li&gt;&#10;&lt;li&gt;A handover flow: the model emits a marker, the outbound pipeline strips it, and the conversation is sent to the advisor API.&lt;/li&gt;&#10;&lt;li&gt;HMAC-signed webhooks and a small test harness.&lt;/li&gt;&#10;&lt;/ul&gt;&#10;&lt;h2 id="why-it-is-interesting"&gt;Why It Is Interesting&lt;/h2&gt;&#10;&lt;p&gt;The hard part is not the prompt. It is the plumbing: keeping a conversation stateful, handling provider failures, and making handover reliable when the model decides it is time to stop talking. The adapter and pipeline boundaries keep the WhatsApp, AI, and memory layers independent.&lt;/p&gt;</description></item><item><title>PrayerTime Calculator</title><link>https://hexrogue.github.io/projects/prayertime/</link><pubDate>Sun, 20 Sep 2026 00:00:00 +0800</pubDate><author>ikanlemas0@gmail.com (Muhammad Nuh)</author><guid>https://hexrogue.github.io/projects/prayertime/</guid><description>&lt;p&gt;Open source: &lt;a href="https://github.com/hexrogue/PrayerTime"&gt;github.com/hexrogue/PrayerTime&lt;/a&gt;.&lt;/p&gt;&#10;&lt;p&gt;A terminal-based prayer time calculator for Malaysia. It computes every time from scratch using spherical trigonometry. No APIs, no network calls, no account, no &amp;ldquo;please enable location services&amp;rdquo; popup.&lt;/p&gt;&#10;&lt;h2 id="the-problem"&gt;The Problem&lt;/h2&gt;&#10;&lt;p&gt;Prayer apps that need the internet are useless in a basement car park, a rural area, or a flight with no wifi. The Sun does not need an API. The calculation is deterministic, and the data is small enough to ship with the binary.&lt;/p&gt;</description></item><item><title>Xray Workers</title><link>https://hexrogue.github.io/projects/xray-workers/</link><pubDate>Sun, 20 Sep 2026 00:00:00 +0800</pubDate><author>ikanlemas0@gmail.com (Muhammad Nuh)</author><guid>https://hexrogue.github.io/projects/xray-workers/</guid><description>&lt;p&gt;Open source: &lt;a href="https://github.com/hexrogue/xray-worker"&gt;github.com/hexrogue/xray-worker&lt;/a&gt;.&lt;/p&gt;&#10;&lt;p&gt;A VLESS proxy system on Cloudflare Workers with user management backed by D1. Edge-deployed, no servers to manage, no VPS to patch at 2 AM.&lt;/p&gt;&#10;&lt;h2 id="the-problem"&gt;The Problem&lt;/h2&gt;&#10;&lt;p&gt;Personal infrastructure should be boring to operate. A VPS gives you control and then asks for it back in the form of updates, certificates, logs, and an alert at the worst possible time. The edge is a better fit for a small proxy: deploy it, forget it, and let someone else own the server.&lt;/p&gt;</description></item><item><title>City2Coordinates</title><link>https://hexrogue.github.io/projects/city2coordinates/</link><pubDate>Sun, 20 Sep 2026 00:00:00 +0800</pubDate><author>ikanlemas0@gmail.com (Muhammad Nuh)</author><guid>https://hexrogue.github.io/projects/city2coordinates/</guid><description>&lt;p&gt;Open source: &lt;a href="https://github.com/hexrogue/city2coordinates"&gt;github.com/hexrogue/city2coordinates&lt;/a&gt;.&lt;/p&gt;&#10;&lt;p&gt;A small geocoding tool that converts Malaysian cities and districts into latitude and longitude data. It was built as a data preparation pipeline for location-aware projects, including the &lt;a href="https://hexrogue.github.io/projects/prayertime/"&gt;PrayerTime calculator&lt;/a&gt;.&lt;/p&gt;&#10;&lt;h2 id="the-problem"&gt;The Problem&lt;/h2&gt;&#10;&lt;p&gt;The prayer time calculator needs coordinates for every city and district it supports. Typing them by hand is boring, error-prone, and exactly the kind of task that should be a script. The data also needs to be cached, because rate limits are real and network calls are not a personality.&lt;/p&gt;</description></item><item><title>ESP32 Recon Rig</title><link>https://hexrogue.github.io/projects/esp32-recon-rig/</link><pubDate>Sun, 20 Sep 2026 00:00:00 +0800</pubDate><author>ikanlemas0@gmail.com (Muhammad Nuh)</author><guid>https://hexrogue.github.io/projects/esp32-recon-rig/</guid><description>&lt;p&gt;The long version lives in the blog series: &lt;a href="https://hexrogue.github.io/blog/esp32-recon-rig-driver-said-no/"&gt;the driver fight&lt;/a&gt; and &lt;a href="https://hexrogue.github.io/blog/esp32-recon-rig-build-guide/"&gt;the build guide&lt;/a&gt;. This page is the project summary.&lt;/p&gt;&#10;&lt;h2 id="the-problem"&gt;The Problem&lt;/h2&gt;&#10;&lt;p&gt;Wifi and Bluetooth analysis usually starts with a laptop, a compatible adapter, and a stack of tools that assume you already know what you are looking at. I wanted to understand the layers from the radio up, on hardware cheap enough to keep on the desk and small enough to carry.&lt;/p&gt;</description></item></channel></rss>