Nanda, backend engineer

I build Go and Node.js backends, native iOS apps in SwiftUI, and products on top of AI models.

Play the projects Contact

Next.js + Claude API

firstep-two.vercel.app

FirStep landing page: simulate your dream career before you regret it

SwiftUI + Go

LensLift showing 650 kcal for a plate of nasi goreng, analysed by AI

AI on TikTok

FirStep career simulation result: Backend Specialist to Tech Lead, 72% probability

About me

I'm Nanda, an Informatics Engineering student at Telkom University in Bandung. I design APIs and data models in Go and Node.js, ship native iOS apps in SwiftUI, put AI models to work inside real products, and talk about AI on TikTok.

Nanda in front of the FirStep screen at the Festival AI Nusantara booth — Presenting FirStep at Festival AI Nusantara.

I like owning a product end to end: the schema, the API contract, the client that calls it and the demo day where people try it. Lately that has meant a national AI showcase, a GEMASTIK paper and my first SwiftUI app.

Focus: API design, data modelling and the apps built on top
Studying: Informatics Engineering at Telkom University
Now: Software Engineering x AI bootcamp at RevoU
Also: Content creator sharing AI tools and tips on TikTok @voynann on TikTok (opens in a new tab)

Projects you can play with

Six things I built. Each one has a working piece right here: live sites, a prototype, an API console and a game.

2026Team lead, full-stack developer

FirStep

AI career simulator for Indonesian students

firstep-two.vercel.app

The full FirStep landing page: hero, stats, the twelve career tools, testimonials and footer

Captured from the live site. Scroll inside the window.

Maps a five-year career path with salary projections, then adds a CV Roaster, a Reality Check score and a First Income guide. Built with Next.js and PostgreSQL, with Claude doing the reasoning.

Team of 3

Finalist at the Focus Target x Microsoft Elevate AI Showcase, exhibited live at FX Sudirman, Jakarta on 5 June 2026.

Next.js
Claude API
PostgreSQL
Vercel

Live demo

2026Solo developer, product and UI

fotokita.space

A photobox in the browser: two peace signs take the shot

Hold up a peace sign with both hands and MediaPipe hand tracking fires the shutter, four times. Then pick one of nine layouts, from a 1x4 strip to a newspaper page, add a cover, filter and frame, and save a print-ready PNG or a boomerang clip for Stories, TikTok and Reels.

Print a layout

Real renders from the app, made with sample photos.

fotokita.space

The camera view after a two-handed peace sign fired shot 1 of 4

Camera: captured from the live site with sample photos.

Solo project, live and public

Live and used by real people to make photostrips with friends and partners. No install and no upload: the photos, the collage and the video are all made on the device.

Next.js
MediaPipe
Canvas API
Tailwind CSS
Vercel

Live site

LENSLIFT

YOUR GYM. TRACKED.

Your AI-powered
gym companion.

Rebuilt for the web from the SwiftUI flow. Tap through it.

2026iOS and backend developer

LensLift

iOS fitness app that reads your meal from a photo

Photograph a meal and Claude Vision (claude-haiku-4-5) returns calories and macros. It also logs workouts, body weight and live gym sessions. SwiftUI on the phone, a Go and Gin REST API with PostgreSQL behind it.

Solo project

My first Swift project, built from zero: a native client and its Go backend, in two repositories.

Swift
SwiftUI
Go
Gin
GORM
PostgreSQL
JWT
Claude API

iOS repo Backend repo

Screens from the real app

2026Backend developer

Olahin Dong

Recipes from whatever is already in your fridge

Finds recipes from the ingredients you have, tracks pantry stock and expiry dates, keeps a monthly grocery budget and runs weekly cooking challenges. I built the whole Express API; two teammates built the Flutter app on top of it.

No token

What's in your fridge?

GET/api/recipes/search?ingredients=telur,nasi

Press Send to call the route.

Runs in your browser with sample data, using the same routes, status codes and messages as the Express API.

Try a recipe search, then call getMe without a token, log in and call it again.

GDG on Campus Telkom University, team of 3

Sole backend developer: I designed the schema, the REST contract and the auth that both Flutter developers shipped against.

Express
PostgreSQL
Prisma
Zod
JWT
REST API

Backend repo

2025Web developer, UI designer

Smoothies Sultan

Scroll-driven landing page for a smoothie brand

A loading reveal, a cup that bursts into fruit as you scroll, a manifesto that lights up word by word and a bento showcase. Built with Next.js, GSAP and Framer Motion.

Solo project

Live on Vercel. It is where I learned to turn a brand direction into motion without losing performance.

Next.js
GSAP
Framer Motion
Tailwind CSS
Vercel

Live site

smoothies-sultan.vercel.app

A smoothie cup bursting into strawberries, mango and blueberries

2026Team lead, ML engineer

GEMASTIK XVIII

Sentiment analysis of Indonesian e-government apps

An NLP pipeline over reviews of M-Pajak, Mobile JKN, MyPertamina and SIGNAL: IndoBERT to label the data, TF-IDF features and an SVM classifier, written up as an IEEE-format paper in Indonesian.

Game

Beat the Model

Twelve app reviews in Indonesian. Label each one negatif or positif. Our SVM scored 91.48% on 775 real test reviews. Can you do better?

The reviews here are written for the game in the style of the dataset.

Drag the card, tap the buttons or use the arrow keys.

Word clouds of complaints per app, dominated by aplikasi, gagal, login, verifikasi, daftar, barcode and bayar — Most frequent words in complaints, one cloud per app. From the team's notebook.

SVM on 775 test reviews

709 right, 66 wrong.

Confusion matrix: rows are the actual label, columns the predicted label.
	Predicted negatif	Predicted positif
Actual negatif	356	31
Actual positif	35	353

Same TF-IDF features, three classifiers

SVM: 91.48%
Naive Bayes: 91.10%
Random Forest: 90.32%

Share of negative reviews

M-Pajak stands out. The paper ties it to complex tax reporting and login problems.

M-Pajak87.3%
Mobile JKN41.7%
MyPertamina37%
SIGNAL32.8%

National competition team

Passed Telkom University's internal selection for GEMASTIK XVIII, Data Mining category. The SVM reached 91.48% accuracy on 775 test reviews.

Python
IndoBERT
scikit-learn
TF-IDF
SVM

Take a break

A one-button runner about getting code to production past the bugs. Reach 500 to ship.

Best 00000Score 00000

Ship It

Jump the bugs and merge conflicts, stay low under the 500s, grab coffee. Hold to jump higher.Tap to jump over bugs. Hold to jump higher.

The toolkit

Every key is a tool I ship with. Press one to see where it was used.

Backend

iOS

AI / ML

Frontend

Backend

Go

Used in LensLift.

The path so far

2026 / Achievement

Seleksi Internal GEMASTIK XVIII, Data Mining

Lolos seleksi internal Telkom University untuk GEMASTIK XVIII kategori Data Mining sebagai ketua tim, dengan riset analisis sentimen aplikasi e-government memakai IndoBERT dan SVM.

Let's build something.

Have a project in mind or just want to say hi? Drop me a message.

gdenand2020@gmail.com

Press it.

Nanda, backend engineerNanda