Gandhi Ert
Julio

Information Technology Undergraduate

About

Information Technology undergraduate at Institut Teknologi Sepuluh Nopember, Surabaya. The work so far sits between data systems, models, and the screen in front of them.

In the GoTo Dicoding cohort I built VendorAI, a vendor recommender for small businesses: a React and Tailwind interface, a FastAPI backend, and PostgreSQL underneath, with a generated explanation for each ranked vendor. Before that I interned as a quality engineer at Telkom Infrastruktur Indonesia, checking network-asset records against what was actually in the field.

I also built a Steam game lakehouse (Kafka, PySpark, MinIO, then a Streamlit dashboard), contributed to Licky, an Android app that reads tongue photos on-device with TensorFlow Lite, and fine-tuned DistilBERT for multilabel emotion classification.

Interested in data systems, machine learning, and how a model shows up in a product.

Experience

Full-Stack Developer Cohort — Dicoding

Aug 2026 — Present · Surabaya

Built web prototypes for a Vendor Recommendation and Procurement Assistant for small businesses, using React, JavaScript, and Tailwind CSS. The ranking engine scores vendors on price, quality, and lead time, and a public search API brings in candidates that are not in the verified set yet. Checked the path from the inference API to the assistant in Postman, kept the repo on Git with branching, and deployed a responsive dashboard on Vercel.

  • React
  • JavaScript
  • Tailwind CSS
  • Git
  • Vercel

Quality Engineer Intern — Telkom Infrastruktur Indonesia

Dec 2025 — Feb 2026 · Jakarta Timur

Profiled and verified network-infrastructure data on a regular schedule so the records stayed consistent and accurate. Updated asset notes, including connected nodes and distribution tracking, until they matched the field, and reconciled gaps between the internal management system and day-to-day operations. With the management team, compressed daily distribution data and tightened quality control.

Projects

VendorAI Recommender

2026

Full-stack recommender where a user submits needs and a budget, then gets ranked vendors with a generated explanation of each pick.

  • React
  • FastAPI
  • Tailwind CSS
  • PostgreSQL

Quantum Phishing Detection

2026

Quantum support vector machine for phishing sites, using a ZZFeatureMap and a fidelity kernel on a 2-qubit simulated circuit. Test accuracy was 55% on a downsampled set, limited by classical simulation.

  • Python
  • Qiskit
  • QSVC

Disguised File Extension

2026

Analyzed an E01 disk image to find files with disguised extensions, including ZIP archives masquerading as JPG, without altering the original evidence. Group assignment.

  • E01
  • Digital Forensics

Steam Data Lakehouse & Game Clustering

2025

Steam game lakehouse with a Kafka streaming pipeline, PySpark ETL, and zone-based MinIO storage. Clustered games with K-Means (k=13) and showed the results in a Streamlit dashboard and a recommendation frontend.

  • Python
  • Apache Spark
  • Kafka
  • MinIO
  • Streamlit

Licky

2025

Native Android app that reads a tongue photo on-device with TensorFlow Lite. I handled CameraX capture inside an MVVM setup with Jetpack Navigation and Room, in a team of 8.

  • Kotlin
  • TensorFlow Lite
  • CameraX
  • Room
  • MVVM

Multilabel Emotion Classifier

2025

Fine-tuned DistilBERT for 14-class multilabel emotion classification with class-weighted loss and a custom head. Macro F1 on the Kaggle test set was 0.42, with a 0.006 gap to validation.

  • Python
  • PyTorch
  • Transformers
  • DistilBERT

Basic Media Schooling

2024

Website for Basic Media Schooling, built with Tailwind CSS and deployed on Vercel through Git.

  • Tailwind CSS
  • Vercel
  • Git

FUSE Filesystem

2024

Custom FUSE filesystem in C that watermarks JPEG files with ImageMagick, reverses prefixed text files, and manages script permissions. Group practicum.

  • C
  • FUSE
  • Linux
  • ImageMagick