Aldo Fahrezy
Résumé
WorkAboutSkillsContactRésumé ↗

Work

Selected projects

From competition ML notebooks to production systems — each written up with the reasoning behind it, the measured results, and honest limitations.

Interpretable ML

Carbon Footprint Prediction — A Comparative, Interpretable ML Study

Nine regressors benchmarked to predict an individual's carbon emissions from behavior — where disciplined preprocessing turned a middling model into a CatBoost regressor with MAE 66.4 (44% below the runner-up), and SHAP explained why.

  • CatBoost
  • LightGBM
  • XGBoost
  • SHAP
  • scikit-learn
  • Pandas
0.9920R² · CatBoost (best of 9)

Technical report · Universitas Indonesia

View case study

Agentic AI

WarungFlow — Autonomous Finance Ops for Micro-Businesses

An LLM-agent system that reads a warung's WhatsApp-style orders, reconciles them against QRIS and bank transfers with fuzzy name-matching, and reports cashflow and a business-health score — winner of Best Payment Use Case at OpenClaw Agenthon Indonesia 2026.

  • Python
  • LLM Agents
  • Tool Use
  • DOKU
  • QRIS
  • Fintech
🏆Best Payment Use Case · OpenClaw 2026

OpenClaw Agenthon Indonesia 2026 · 🏆 Best Payment Use Case

View case study

Geospatial ML

BanjirWarga — Community Flood Intelligence

A flood-intelligence platform for Indonesian cities: a medallion data pipeline feeding an explainable 0–100 risk score, a report trust score, and a Dijkstra-based safe-route engine over a risk-weighted road graph — served through a PostGIS backend and an offline-capable map PWA.

  • FastAPI
  • PostGIS
  • PostgreSQL
  • React
  • TypeScript
  • Leaflet
0–100Explainable flood-risk score

GEMASTIK PPL 2026

View case study

Multimodal ML

TexTell — Multimodal Fashion Tagging & Listing

A mobile app that turns a photo of a garment into structured attributes and a ready-to-post shop listing — backed by two models I trained on Kaggle and exported for inference: a PyTorch vision tagger and a LoRA-fine-tuned text generator.

  • PyTorch
  • LoRA
  • Computer Vision
  • NLP
  • FastAPI
  • React Native
CV + NLPTwo trained models, one product

Multimodal ML product

View case study

Tabular ML

Speed-Dating Match Prediction — Satria Data BDC 2025

Predicting mutual 'matches' from speed-dating data by engineering pairwise features between two people, then benchmarking four models under honest 5-fold cross-validation — and resisting the in-sample number that looked too good to be true.

  • scikit-learn
  • Random Forest
  • XGBoost
  • CatBoost
  • Feature Engineering
0.798OOF ROC-AUC · Random Forest

Satria Data Big Data Challenge 2025 · UI selection

View case study

Tabular ML

Loan-Default Prediction at Scale

A textbook-clean gradient-boosting pipeline over ~850k loan applications — feature engineering, five-fold out-of-fold cross-validation, and leakage-free categorical handling — reaching an OOF ROC-AUC of 0.923.

  • CatBoost
  • scikit-learn
  • Feature Engineering
  • Cross-Validation
0.9227Out-of-fold ROC-AUC

Kaggle Playground Series (S5E11)

View case study

Distributed Systems

BidMart — Distributed Auction Marketplace

A production-shaped microservices marketplace whose Rust auction engine settles money correctly even when processes crash mid-transaction — via a transactional outbox, a lease-locked job queue, and a two-phase settlement saga.

  • Rust
  • axum
  • gRPC
  • RabbitMQ
  • PostgreSQL
  • Spring Boot
6Bounded-context services

Advanced Programming project · Universitas Indonesia

View case study

Full-Stack

LigaPass — Cross-Platform Football Ticketing Ecosystem

A football ticketing, news, and reviews platform spanning a Django SSR web app and a Flutter mobile client — with real multi-rail Midtrans payments, server-verified Google OAuth, and a resilient three-tier match-data pipeline. Deployed, with signed-APK CI/CD.

  • Django
  • Flutter
  • PostgreSQL
  • Midtrans
  • Google OAuth
  • GitHub Actions
2Clients: SSR web + Flutter

Platform-Based Programming · Universitas Indonesia

View case study

Also competed in

  • Data Slayer 3.0Video drowsiness detection — MediaPipe FaceMesh features + gradient boosting.
  • Objective Quest 2025Sentence-duration regression over 23k Indonesian court decisions (NLP).
  • GEMASTIK 2025 · Data MiningCross-lingual retrieval with neural MT + anagram fingerprinting.
  • Datavers Anava5-class ride-hailing trip-risk classification with physics-based features.
  • MineToday · IT Today 2025Student-outcome prediction from multi-source academic data.
  • Gammafest IPB 2025Scientific-paper citation link prediction.