Bekasi, Indonesia
Muhammad Faza Pinaringgi
AI Engineer who builds LLM-powered features and AI agents for real users. Master's student in Computer Science (Data Science) at BINUS University.
Experience
AI Engineer & Full-stack Web Developer
Jul 2026 – PresentPT OOH Kupu Kupu Global (OOH Group) · Sole engineer for four web products
SemarLanguage · semarlanguage.com
- Took an AI-scored language-test platform from concept to a paid service (Next.js, TypeScript, Supabase, Vercel).
- Designed the IELTS scoring engine: Claude scores Writing against the official criteria, Whisper transcribes Speaking, and Listening/Reading are scored by rules.
- Calibrated the Writing scorer on 46 band-scored reference essays (mean bias of −0.15 band).
ooh.co.id, jobpmi.co.id, oohmiracle.au
- Took over and rebuilt the hosting of inherited Laravel sites, upgraded them to Laravel 12, and cleared all 48 known security advisories on ooh.co.id.
- Built lead scoring, invoicing, payment integration, and single sign-on to SemarLanguage; migrated a production database with zero row mismatches.
Projects
Simpul: AI Data-Hub Prototype with Gemini Agents
202610-day build for a team in an internal corporate AI competition; the team won 1st place.
- React/TypeScript prototype with serverless Gemini endpoints; the analyst agent answers through function calling over 14 query tools, so every number is computed by code rather than generated.
- On 8 messy distributor files (simulated data), it recognised every file type and automatically mapped 91% of columns.
Master's Thesis: Uncertainty-Aware Multi-Teacher Knowledge Distillation
2025 – Present- Trained three teacher models (DenseNet-121, ResNet-50, RAD-DINO ViT) on 112,120 NIH chest X-rays; the best reaches mAUC 0.840, on par with CheXNet (0.841).
- Distilling them into a lightweight EfficientNet-B0 student that weights each teacher by its uncertainty.
Financial News Sentiment Analysis
2025- Fine-tuned and compared RoBERTa, FinBERT, and BiLSTM for 3-class financial news sentiment; general-purpose RoBERTa (82.8% accuracy) beat domain-specific FinBERT (81.0%).
Plant Disease Detection
2024- Built and deployed a Gradio web app where users upload a leaf photo and get one of 38 plant-health classes with a confidence score, served by a TensorFlow model on Google Cloud.
Publications
- Dzaky, J. D., Pinaringgi, M. F., Purnomo, F., & Prasetyo, A. (2025). 3D Dynamic Flood Visualization System with AI-based Prediction Using LSTM, ANUGA and Leafmap for Jakarta Urban Area. ICCSCI 2025, Procedia Computer Science (Elsevier).
- Wijaya, M. G., Pinaringgi, M. F., Zakiyyah, A. Y., & Meiliana (2024). Comparative Analysis of Machine Learning Algorithms and Data Balancing Techniques for Credit Card Fraud Detection. ICCSCI 2024, Procedia Computer Science (Elsevier). Oral presenter.
Education & Certification
Master of Computer Science, Data Science Specialization
2025 – PresentBina Nusantara University · Fast-track program · GPA 3.93 / 4.00
Bachelor of Computer Science, Intelligent Systems (AI) Specialization
2022 – 2026Bina Nusantara University · S.Kom. · GPA 3.59 / 4.00
Microsoft Applied Skills: Generate reports with AI research agents
Nov 2026Microsoft · hands-on lab assessment · Verify credential
Skills
AI / LLM: AI agents (function calling, tool use), prompt engineering, LLM evaluation, Claude API, Gemini, Whisper
Machine learning: PyTorch, TensorFlow/Keras, Hugging Face Transformers, scikit-learn; CNNs, Vision Transformers, LSTM
Software: Python, TypeScript, PHP, SQL; Next.js, React, Laravel; PostgreSQL (Supabase), MySQL
Delivery: Git, GitHub Actions CI, Vercel, Linux servers, Gradio
Contact
Open to AI engineering and AI automation roles.