Building practical AI systems, modern ML workflows, and deployable intelligent products.
📍 Jakarta, Indonesia · 🇮🇩 Indonesian · 🌸 Learning Japanese
Novia is an AI assistant and coding companion exploring conversational AI, local inference, multilingual interaction, and voice interfaces.
- AI agent workflows and coding assistance
- Indonesian and Japanese interaction
- LLM experimentation and local model inference
- Voice-first assistant experience
Explore Novia model experiments on Hugging Face.
Novia is an ongoing R&D project. Features and architecture evolve as experiments are validated.
| Area | Tools and focus |
|---|---|
| Programming | Python, SQL, Bash |
| Machine Learning | Scikit-learn, model training, evaluation, feature engineering |
| Deep Learning | PyTorch, TensorFlow, neural networks |
| Generative AI | LLMs, prompt engineering, RAG concepts, Hugging Face, Ollama |
| Model serving | FastAPI, REST APIs, inference services |
| MLOps | MLflow, experiment tracking, model versioning, reproducible pipelines |
| Infrastructure | Docker, Kubernetes / K3s, Linux, GitHub Actions |
| Cloud & design | AWS fundamentals, cloud deployment, Figma, UI/UX |
DATA → TRAIN → EVALUATE → TRACK → DEPLOY → MONITOR
I enjoy turning experiments into maintainable systems: reproducible training, measurable evaluation, reliable inference, and observable deployments.
- Building practical foundations in machine learning and deep learning.
- Exploring LLM application patterns, agents, and local inference.
- Developing repeatable model training and deployment workflows.
- Learning how to monitor and operate ML services with MLOps practices.
- Improving product experience through thoughtful UI/UX and Figma-inspired design.
| Project / area | Focus |
|---|---|
| Novia | AI assistant, coding buddy, multilingual and voice interface |
| LLM experiments | Local inference, model evaluation, and fine-tuning exploration |
| ML learning lab | Python notebooks, training experiments, and evaluation |
| MLOps lab | Containerized inference, deployment workflows, and monitoring concepts |