Institut Teknologi Sepuluh Nopember
Information Systems — Master’s
- Developing LLMs for real-world applications in specific domains
- Learning how to optimize LLMs
- Building scalable LLM systems
Moh. Rosy Haqqy Aminy
PORTOFOLIO
“No Royal Road In Geometry”
AI Engineer and Master’s student in Information Systems at Institut Teknologi Sepuluh Nopember, working where model capability meets model reliability — LLM alignment, retrieval-augmented generation, and applied machine learning.
- 5Publications
- 3IEEE conferences
- 3Registered HAKI
- 3.91Current GPA
01 — Who I am
About Me
I am Rosy Haqqy, a Master’s student in Information Systems at Institut Teknologi Sepuluh Nopember. My research is on large language models for high-stakes domains, covering alignment pipelines from supervised fine-tuning to preference optimisation with LoRA, and retrieval systems that combine dense and sparse search with knowledge graphs. I am most interested in where model capability meets model reliability.
Previously I worked as an AI Engineer at Bank Rakyat Indonesia on retrieval-augmented chatbots. At ITS I led the AI team for SENOPATI, the university’s self-hosted RAG assistant, and I coordinate research at the ADDI Lab. I have five publications and have presented at three IEEE international conferences.
A background in mathematics and competitive programming shaped how I work: formalise first, then build. I am passionate about how AI and formal logic can work together to create impactful solutions, and I aspire to contribute to future innovations at Google DeepMind.
Focus areas
Tooling
02 — Background
Education
-
-
Institut Teknologi Sepuluh Nopember
Informatics Engineering — Bachelor’s
- Studied artificial intelligence
- Studied algorithm design and optimisation
- Developed research in machine learning and data hiding
03 — Project & Achievement
Portfolio
Four projects where I led or built the machine-learning core.
2025
SENOPATI
AI Team Leader
SENOPATI (Sepuluh Nopember Personal Assistant with Intelligence) is a generative AI chatbot created at Institut Teknologi Sepuluh Nopember. It is a locally hosted assistant using a Retrieval-Augmented Generation approach built atop a self-hosted LLM powered by tools like Ollama and OpenWebUI, allowing it to run entirely offline for enhanced data privacy and cost control. It ingests user-uploaded documents, retrieves relevant information, and generates context-aware responses — enabling accurate, real-time assistance in education.
What I learned: how to build a RAG system on a locally hosted LLM, and how to lead and manage a large, multidisciplinary team of fellow students to deliver a scalable, privacy-focused AI solution.
- Soft skills
- Teamwork · Critical thinking · Problem solving · Leadership
- Hard skills
- FastAPI · LLM · RAG
2025
BrainAI
Team Leader
An improved version of the BrainHealth app — an application that detects brain tumours down to their stage, using a Vision-Language Model together with an LLM to power the chatbot and the recommendation system.
What I learned: how to evolve an existing product and work with Vision-Language Models (VLM).
- Soft skills
- Teamwork · Critical thinking · Effective communication
- Hard skills
- LLM · VLM · Python
2024
BrainHealth
Machine Learning · Backend · DevOps
An application that assists doctors by automating the early detection of tumour types and providing treatment recommendations for patients according to the tumour type detected.
What I learned: how to build a brain-tumour prediction model and deploy it to production.
- Soft skills
- Teamwork · Critical thinking · Effective communication
- Hard skills
- Flask · Docker · TensorFlow
2024
WisYuk
Machine Learning Team Leader
WisYuk is a platform for the Indonesian tourism industry offering a comprehensive solution that makes travel more convenient and memorable. It uses NLP and a recommender system to surface the best travel recommendations based on trends and Twitter review sentiment.
What I learned: how to build a recommendation model using NLP principles to analyse trends and sentiment for personalised travel suggestions.
- Soft skills
- Teamwork · Critical thinking · Effective communication
- Hard skills
- TensorFlow · NLTK · Flask · BERT
04 — Industry
Working Experience
Bank Rakyat Indonesia
AI Engineer
BRILiaN Internship Program in the Enterprise Data Management Analytics Division, working on retrieval-augmented chatbots. Completed with an Excellent score.
PT Pertamina EP Cepu
Data Analyst
Practical work program (PKL) in Facility & Maintenance — Zona 12, certified by the Human Capital Business Partner division.
05 — Research
Publications
Five papers across large language models, steganography and reversible data hiding.
-
01
MapReduce and K-Means Clustering Method for Long Text Summarization on Large Language Model
-
02
Hiding Secret Data Using Reduced Differences within the Pixels of an Image
-
03
MedicalFuzzySec: a novel steganography technique using fuzzy logic to secure electronic patient data (EPD) concealment in medical images
-
04
Towards a high-capacity data concealment for spatial domain image-steganography
-
05
A reversible data hiding approach in medical images using difference expansion
06 — Other Experience
Achievements
Conference talks
Oct 2025 · Malaysia
IEEE Region 10 Conference (TENCON 2025)
Speaker — “MapReduce and K-Means Clustering Method for Long Text Summarization on Large Language Model”, Sabah International Convention Centre, Kota Kinabalu.
Dec 2023 · Thailand
IEEE Conference on Computational Intelligence and Communication Networks (CICN 2023)
Speaker — “A Reversible Data Hiding Approach in Medical Images Using Difference Expansion”, Bangkok.
Dec 2023 · India
IEEE 3rd Conference on Mobile Networks and Wireless Communications (ICMNWC 2023)
Speaker — “Towards a High-capacity Data Concealment for Spatial Domain Image-steganography”, Tumakuru.
Registered intellectual property (HAKI)
Steganography
Steganografi: Ekspansi Diferensial Dalam Pengelompokan Baris 2-Pixel Pada Citra Medis Untuk Meningkatkan Kualitas Gambar Stego
View certificate
Medical imaging
BrainHealth: Aplikasi Mobile Automasi Pendeteksi Tipe Tumor Otak Pada Citra Magnetic Resonance Image Menggunakan Convolutional Neural Network
View certificate
Platform
InBengkel: Platform Digital Two Sided Users Yang Mempertemukan Bengkel Dengan Pengguna Kendaraan Bermotor Secara Cepat Dan Tepat
View certificate
Competition
Sep 2023 · National
Gemastik 2023 — Finalist, Programming Division
National finalist in the Programming (Pemrograman) division of Gemastik 2023, held by Balai Pengembangan Talenta Indonesia with Universitas Brawijaya in Malang, 12–16 September 2023.
Speaking & workshops
Mar 2026
Workshop NVIDIA 2026
Speaker — “Upgrade your skills with ADDI Lab”: hands-on AI supercomputing, faster model training and HPC.
Nov 2025
AXION 2025 — Avalon Expo of Innovation
Talkshow guest star as IRISA AI Engineer — “Membentuk Generasi Talenta AI Indonesia”, Balai Pemuda Surabaya.
Jun 2024
Bangkit 2024 Batch 2 webinar
Student representative for the Machine Learning cohort in the “Kepoin Program” free webinar.
May 2024
Programming Division workshop
Speaker — “How to become the competitive programming champion at Gemastik 2024?”, with Tim Euler.
Certifications
- Google IT Automation with Python
- Learn Data Analysis with Python
- Mathematics for Machine Learning and Data Science
- Machine Learning Specialization
- DeepLearning.AI TensorFlow Developer Professional Certificate
- Structuring Machine Learning Projects
- TensorFlow: Data and Deployment Specialization
- TensorFlow: Advanced Technique Specialization
- Generative AI for Everyone
- AWS Academy Machine Learning Foundations
- AWS Academy Cloud Operations
- AWS Academy Cloud Foundations
07 — Contact
Thank you!
Let’s connect — I am open to research collaboration, AI engineering roles and speaking invitations.