Agent Platform
Siberflow
Multi-provider AI platform with streaming tool calling, file and database access, task checklist support, session persistence, and CLI plus VS Code delivery.
- TypeScript
- Agent Loop
- VS Code
- CLI
20+
Delivered software projects across AI and non-AI work
5+
AI-focused projects and labs
4
Main delivery surfaces: CLI, VS Code, API, Web
Selected Work
Projects that define the portfolio
Developer Assistant
IdSiberAi-CLI
Command-line AI assistant with multi-provider support, storage and database tooling, provider fallback, and practical system automation features.
- JavaScript
- LLM Integration
- S3
- Automation
Coding Agent
IdSiberCoder-VSCode
VS Code extension that lets an AI developer agent inspect files, search workspace context, run commands, and edit project content through tool calling.
- VS Code API
- Tool Registry
- Parallel Tools
Document Intelligence
information_extraction
OCR workflow for Indonesian identity cards and passports using OpenCV, Tesseract, EasyOCR, and optional LLM-assisted extraction.
- Python
- OCR
- OpenCV
- Flask
AI/NLP Lab
media_analytics
Experimental Indonesian-language media analysis workspace covering scraping, SQLite datasets, IndoBERT embeddings, clustering, keyword extraction, chatbot experiments, and local LLM runtime.
- IndoBERT
- KeyBERT
- GGUF
- PyTorch
Local Model Runtime
GGUF and llama.cpp experimentation
Hands-on local model experimentation with quantized GGUF models, `llama.cpp`, `llama-cpp-python`, interactive chat scripts, and OpenAI-style local API wrapping.
- GGUF
- llama.cpp
- llama-cpp-python
- FastAPI
Experience Themes
Where the engineering depth shows up
AI Agents
Building systems where models do more than respond: they inspect context, choose tools, execute actions, and continue a task loop.
Multi-Provider LLMs
Designing product flows that work across multiple providers with switching, fallback, and interface consistency.
OCR and Extraction
Combining OCR engines, preprocessing, heuristics, and optional LLM assistance to structure noisy document data.
Indonesian NLP
Exploring IndoBERT, TF-IDF, clustering, stemming, keyword extraction, and question-answering workflows for local-language data analysis.
Local Model Serving
Running and wrapping GGUF models locally with `llama.cpp` and `llama-cpp-python`, including chat endpoints and Apple Silicon oriented experimentation.
Institutions
Delivery contexts and organizations reflected in the portfolio
Kementerian Perdagangan
Kominfo RI
Pemerintah Kota Yogyakarta
DLHK Provinsi DIY
DINPMPTSP Kabupaten Purworejo
Kota Samarinda
PT PP Persero
Badan Intelijen Negara
AKS-AKK Yogyakarta
Broader Delivery
Non-AI experience that strengthens the AI profile
Flutter Mobile Apps
Delivery experience on mobile applications for institutional and government workflows, including internal service applications.
GIS Platforms
Built geoportals, interactive maps, stock maps, and spatial monitoring systems using LeafletJS, GeoServer, and relational databases.
Operational Systems
Worked on recruitment portals, CBT systems, licensing systems, attendance systems, tax-payment systems, and monitoring dashboards.
CMS and Dashboards
Delivered admin interfaces, content management systems, and reporting tools for public-facing and internal institutional use.
Non-AI Highlights
Selected software delivery outside the AI track
Flutter Mobile
E-Kemendag, G20 TIIWG, Hero Mobile
Institutional mobile application delivery for internal services, information access, and support workflows in government environments.
- Flutter
- Mobile Apps
- Government
GIS Systems
Geoportal and interactive map platforms
Geospatial products for public information, stock monitoring, environmental data, Covid monitoring, and map-based city services.
- LeafletJS
- GeoServer
- PostgreSQL
Operational Workflows
Licensing, recruitment, CBT, attendance, tax systems
Business-critical systems designed for institutional process management, administrative control, and service workflows.
- Laravel
- CodeIgniter
- SQL Server
- MySQL
CMS and Dashboards
Portal management and reporting platforms
Admin systems, dashboards, and CMS workflows for publishing, monitoring, reporting, and decision support in organizational settings.
- CMS
- Dashboards
- Reporting
Combined Stack
Technology range across AI and software delivery
AI and Data
- OpenAI-compatible providers
- IndoBERT, mT5, KeyBERT
- OCR with Tesseract and EasyOCR
- PyTorch, Transformers, GGUF runtime
- llama.cpp and llama-cpp-python
- Embedding, clustering, and topic analysis
Application Delivery
- TypeScript, JavaScript, Python, PHP
- Flutter, Vue, VS Code Extension API
- FastAPI, Flask, Laravel, CodeIgniter
- PostgreSQL, MySQL, SQLite, OracleDB, SQL Server
- LeafletJS, GeoServer, REST API, CMS, dashboards
Professional Positioning
Best-fit role categories
AI Engineer
LLM Application Developer
AI Tooling Engineer
Full-Stack Developer with AI Automation Focus
The strongest signal across the portfolio is practical AI engineering: building assistants, agents, extraction pipelines, and experimentation environments that connect models to real tools, data, and workflows.
What makes the profile stronger is that this AI work sits on top of broader delivery experience in mobile, GIS, dashboards, CMS, and institutional systems. That combination is well aligned with roles involving developer productivity, LLM applications, workflow automation, document intelligence, and applied NLP.
Contact
Available for AI, automation, and full-stack engineering work
Best fit for roles involving AI tooling, LLM applications, workflow automation, document intelligence, internal platforms, and production software delivery.