Hi, I’m Reyno Benedict.

1. 012026Solo

   Featured project

   ICP Intelligence — Indonesian crude price forecasting platform

   An end-to-end MLOps platform that rebuilds Indonesia’s official crude price (ICP) series and forecasts it alongside global benchmarks.

   • Python
   • FastAPI
   • MLflow
   • React
   • Cloudflare Workers
   Repository for ICP Intelligence (opens in new tab) Staging demo of ICP Intelligence (opens in new tab)
   View detailsHide details for ICP Intelligence

   Problem

   Indonesia’s official monthly crude oil price (ICP) is set by the Ministry of Energy and Mineral Resources (ESDM) and published in decree documents, so building a clean history to analyse and forecast requires extracting it from PDFs.

   Data pipeline

   PDF text extraction (PyMuPDF) and OCR (Tesseract) on ESDM decrees, with each month reconciled across sources and combined with Brent, Dubai, WTI and the USD/IDR rate.

   Forecasting

   1–6 month forecasts with uncertainty ranges; candidate methods compared with rolling-origin backtesting; models tracked in MLflow, data versioned with DVC, and a continuous-training workflow in GitHub Actions.

   Application

   FastAPI backend and a React web app with saved scenarios, rule-based insights, PDF/Excel/CSV reports, user accounts and an operator area for datasets, models and jobs.

   Infrastructure

   Cloudflare Workers gateway with a D1 database, Docker and Docker Compose, Prometheus and Grafana monitoring, and CI that runs linting, type checks, tests and secret scanning.

   My role

   Designed and built the whole system as a solo project; it began as a course MLOps assignment and grew into a full platform.

   Status

   Portfolio project with a public staging deployment — not a production service. The staging demo’s forecasting features work only while the self-hosted backend is online. Model evaluation is ongoing.

2. 022026Team of 3Main developer

   Business record matching for BPS Kota Malang

   An internal web tool that matches records of the same business across datasets, then routes candidate pairs to staff for review.

   • TypeScript
   • Next.js
   • FastAPI
   • PostgreSQL
   • FAISS
   Repository for business record matching (opens in new tab)
   View detailsHide details for business record matching

   Problem

   Business records from different datasets need to be linked when they refer to the same business, and every proposed match has to be checked by staff.

   Contribution

   Main developer on a three-person team.

   Approach

   A Python FastAPI service normalises text and generates candidate pairs using character n-gram TF-IDF, truncated SVD, FAISS nearest-neighbour search and RapidFuzz scoring. A Next.js and PostgreSQL app provides role-based workflows: admins upload datasets and run matching jobs, staff verify pairs as match, non-match or review, and the office head monitors progress — with audit logs and CSV export.

   Status

   Work-in-progress MVP. The real datasets are restricted, so the repository uses anonymised data, and no accuracy figures are reported because there are no ground-truth labels.

3. 032026SoloCourse project

   Indonesian job market analysis with Apache Spark

   Scraped job listings and Spark-based analysis of Indonesia’s job market.

   • Python
   • PySpark
   • Spark MLlib
   • Selenium
   Repository for job market analysis (opens in new tab)
   View detailsHide details for job market analysis

   Problem

   Understand hiring patterns in Indonesia — which companies, locations, job types and salary ranges appear in job listings.

   Contribution

   Solo final project for the Big Data Analytics course.

   Approach

   Listings scraped from Glints with Selenium and BeautifulSoup, cleaned with pandas, analysed with Spark SQL, and job titles grouped with a Tokenizer → TF-IDF → K-Means pipeline in Spark MLlib.

   Status

   Completed course project.

All repositories on GitHub (opens in new tab)

• Software Engineering

  Typed, tested applications with clear boundaries between services.

  • TypeScript, React & Next.js
  • FastAPI services
  • Tests & CI with GitHub Actions

• Machine Learning

  Forecasting, similarity search and classification, evaluated against baselines.

  • scikit-learn
  • Time-series backtesting
  • TF-IDF, FAISS & fuzzy matching

• Data & MLOps

  Taking models from notebooks to versioned, monitored pipelines.

  • MLflow & DVC
  • Apache Spark (SQL, MLlib)
  • Prometheus & Grafana

• Infrastructure

  Containers, orchestration and edge deployment, through my projects and coursework labs.

  • Docker & Docker Compose
  • Kubernetes (kubeadm, HPA) in coursework labs
  • Cloudflare Workers & D1