Software Engineer AI Developer Agentic Developer
Software engineer from Universitas Prasetiya Mulya, class of 2026. I build AI agents and automation systems for clients, after 2.5 years as an Agentic Developer at PT Suarana Nusantara Indonesia.
01About
Fig. 01 — Melvern Mogens
I studied Software Engineering at Universitas Prasetiya Mulya (class of 2026, GPA 3.66).
From November 2023 to May 2026 I was an Agentic Developer at PT Suarana Nusantara Indonesia, building AI automation systems and integrating local AI models. Alongside that, I interned at PT Metaverse Indonesia Makmur (Aug–Dec 2025), working on IoT and embedded systems.
Now I build for clients and for myself: a stock-signal app running 24/7 for a trader, an AI research desk a client bought, the system that took my family's rubber business off paper, and a fleet of AI agents that code, design and test with me. Most of it is Python, TypeScript, Flutter, Kotlin and Swift, with LLMs and WhatsApp doing the heavy lifting. Excelsior.
Universitas Prasetiya Mulya
02Skills
Self-rated — how comfortable I am building something real with each one, not how many years it has been on a résumé.
03Experience
Independent · Client work
Agentic AIPythonTypeScriptWhatsApp
PT Suarana Nusantara Indonesia
Agentic AILLMsAutomationPythonWhatsApp
PT Metaverse Indonesia Makmur
IoTEmbedded SystemsNetwork Optimization
Invaders Board Game Station
Customer ServiceLeadershipEvent Management
PT Indah Jaya
Financial ReportingCash FlowExcel
PT Gemilang Hasil Bumi
Public SpeakingNegotiationBranding
04Work
01
Client · Live
A private stock and crypto signal app for a trader client, running 24/7 on his own Mac mini. The main lists only show rules that passed honest backtests, and a radar flags every 10,000-lot trade on the exchange.
PythonFastAPISQLitePreact PWAWhatsApp
02
Client · Sold
A team of AI agents that screens 45 LQ45 stocks and the top-20 crypto every morning against a trader's own rules, then reports on WhatsApp. It learns from his take/skip replies. The client bought it along with a course, and its engine became Sinyal.
PythonMulti-agentpandasWhatsApp
03
Thesis · In use
My thesis: the system that took my family's rubber trading business off paper. A Django REST backend and Flutter app handle receipts, payments, warehouse cash and farmer loans, print Bluetooth receipts, and forecast tomorrow's rubber price with an SVR model.
Django RESTFlutterscikit-learnFirebase
04
Client · Delivered
An internal tool for a property developer's procurement team. Upload contractors' bills of quantities and it lines them up item by item, flags outlier prices with a robust z-score and recommends a vendor. The chat assistant handles uploads but never does the maths itself.
Next.jsReactSQLiteECharts
05
Personal · Daily driver
I run a fleet of specialist AI agents (coder, designer, QA, video) under one orchestrator, and built a native Android client for it. It speaks the desktop app's WebSocket JSON-RPC protocol, so every session streams live on my phone. Now at v28.5.
KotlinJetpack ComposeWebSocketAgentic AI
06
Personal · Working
A local studio for my TikTok and Reels. I record a script section by section in the browser, then an agent pipeline trims, transcribes, adds captions and motion graphics, and renders a 1080×1920 MP4. Every rejected edit needs a reason, so it learns my taste.
TypeScriptHonoReactGSAP
07
Demo · TikTok
A fake fashion brand with a real WhatsApp bot that answers size questions like a human admin and takes orders, holding stock for 30 minutes and logging each order. I built it to demo AI for small business on TikTok.
Node.jsBaileysLLMWhatsApp
08
Live · Play it
A Bookworm-style word RPG for browser and phone: spell words on a 4×4 letter board to hit enemies. It has a 5-chapter adventure, 4 heroes, online Duels, a Daily Challenge and a real-time 90-second Blitz race.
TypeScriptViteSupabasePWA
Visit live site ↗
09
Live · baoseas.site
For my brother's Blox-Fruits-style Roblox game I build new fruits (models, skills, VFX), the 3D ocean-dive website, and Cap'n Bao, the Discord bot. Fan requests from the site land in Discord, and a TikTok bot uses an LLM to pull fruit ideas out of comments.
Next.jsthree.jsSupabaseRoblox Luau
Visit live site ↗
10
macOS · Daily use
I rebuild the Mac utilities I rely on, my way, in Swift: Snapper (CleanShot-style capture, recording and OCR), Kopi (clipboard history), Snag (drag shelf), Switchboard, TuckBar, LimitBar (Touch Bar AI-usage meter) and Tune, my fork of the open-source Kaset YouTube Music app.
SwiftSwiftUIAppKitScreenCaptureKit
11
Android · APK 2.6
An Android app for collectors: enter the weights of the sealed blind boxes on a shop shelf and it solves the whole 12-box case to say which one most likely holds the rare chase figure. It says plainly when weight alone can't tell.
KotlinJetpack ComposePythonData modelling
12
Machine Learning
A licence-plate detector served through Flask. I collected and labelled the data, trained the YOLOv5 model, deployed it as a web service, and kept tuning the accuracy against real traffic footage.
YOLOv5FlaskPython
13
Mobile
A learning app in Flutter with a Django backend: user auth, course tracking and progress, and a responsive UI I designed in Figma first. I handled the whole flow, from the wireframes to the REST API and the mobile client.
FlutterDjangoFigma
14
AI
A voice assistant that listens and talks back: speech-to-text in, text-to-speech out, with RAG so the answers stay grounded in real documents. Everything runs on a local model, so it works offline and keeps data private.
OllamaFlaskCoqui TTSRAG
15
AI
A multimodal assistant that can read images as well as text. It runs on a local LLaVA base model with image RAG on top, so it can answer questions about photos and documents without sending anything to the cloud.
LLaVAFlaskImage RAG
Nothing in this category yet.
05Awards
In2Food 2023 asked teams to turn food waste into something useful. Our team placed second.
06Contact
I’m open to junior and full-time roles, and to freelance AI projects. Email is the fastest way to reach me.
melvernmogens@gmail.comIn2Food 2023, run by Universitas Prasetiya Mulya, challenged teams to design innovative ways to reduce food waste. Our submission took second place.
Read the article