Independent AI Research Lab and Hobbyist. Passionate about building agentic AI systems, training deep learning models, and exploring the frontier of generative AI, large language models (LLM), retrieval-augmented generation (RAG), computer vision AI, LiDAR sensing, and speech-to-text systems. With nearly two decades of background in IT infrastructure, network security, and remote operations, now channeling that experience into AI research, AI engineering, LLM fine-tuning, RAG pipelines, vision AI, and developing intelligent automation solutions.
|
🎤 TranscribeAI🔒 Ideal for processing confidential and financial data. Security-focused, local data processing, NO DATA LEAKED. 100% Local AI Transcription with Speaker Diarization. No API key, no cloud, no cost — runs completely offline on your machine. Powered by dual engine (faster-whisper CPU + mlx-whisper Apple Silicon GPU), supports 99+ languages with auto-detection, speaker identification, and multi-format input/output. Features 5 AI model sizes from tiny (39M) to large-v3 (1.5B), smart progress pipeline, auto model caching, and a professional dark theme web UI.
Python
Whisper
Apple Silicon
100% Offline
MIT License
|
|
📄 AI PDF Tools🔒 Ideal for processing confidential and financial data. Security-focused, local data processing, NO DATA LEAKED. A lightweight, privacy-focused web application providing a comprehensive suite of PDF processing utilities. Convert PDF to Word, compress, OCR with 20+ languages via Tesseract, merge, split, and unlock password-protected PDFs — all processed 100% locally on your machine with zero cloud dependency. Built with Flask backend and vanilla JavaScript frontend, featuring a modern tab-based UI with drag & drop upload, automatic temp file cleanup, REST API with 6 endpoints, and Docker support for containerized deployment.
Flask
Python
Tesseract OCR
pdf2docx
100% Local
Docker
MIT License
|
|
🚗 Tesla Self-Driving AI SimulationReal-time 3D autonomous vehicle simulation built entirely in pure JavaScript Canvas with zero external dependencies. Features a custom perspective projection engine, procedurally generated roads with curves and intersections, AI-controlled traffic vehicles with detection bounding boxes, 360° LiDAR sweep with point cloud visualization, dynamic traffic light state machine, and a cinematic HUD displaying sensor status (8 cameras, 4x LiDAR, radar, 12 ultrasonic), speed, battery, and real-time object detection accuracy for lanes, vehicles, pedestrians, and traffic signs. All in a single HTML file.
JavaScript
HTML5 Canvas
3D Projection
LiDAR Simulation
Zero Dependencies
Vercel
|
|
♞ Knight Chess AIA strategic chess variant with 5 knights per side on an extended 8×9 board, powered by a custom Minimax AI Engine with alpha-beta pruning. The non-standard 8×9 board (vs standard 8×8) combined with 5 knights per side makes it almost impossible for humans or traditional chess AI engines like Stockfish to win — no existing opening book, endgame tablebase, or positional theory applies. The AI searches up to depth 5 with position evaluation covering material balance, piece mobility, king safety, and knight fork detection. Features three difficulty levels (Easy, Medium, Difficult), 10-minute game timer, token economy, and leaderboard system. 3 pawns are randomly replaced by knights each game, creating infinite strategic possibilities.
Minimax AI
Alpha-Beta Pruning
Next.js
TypeScript
Web Worker
Vercel
|
|
🧠 Transformer Explainer (SimulasiLLM)Interactive LLM Attention Simulation that visualizes how GPT-2 style transformer models process and generate text. Explore token embeddings, Q/K/V projections, causal masking, softmax sampling, and attention weights in real-time. Adjust temperature and top-k parameters to see how they affect next-token prediction. Features autoregressive text generation, probability distribution visualization, and a full breakdown of the self-attention mechanism — all running in the browser with zero dependencies and zero frameworks.
Transformer
GPT-2 Simulation
Attention Mechanism
Pure HTML/CSS/JS
Zero Dependencies
|
|
🔥 PyTorch Neural Network SimulationInteractive web-based educational platform for learning PyTorch and neural network fundamentals through real-time visualization. Build, train, and visualize networks live while learning PyTorch with auto-generated code that mirrors your configuration in real-time. Features live neural network simulation with adjustable architecture (layers/neurons), clickable neuron inspection during training, XOR Playground with decision boundary visualization, and a complete training pipeline walkthrough.
PyTorch
Neural Networks
Next.js
TypeScript
Framer Motion
Vercel
|
|
💬 Open Chatbot🔒 Ideal for processing confidential and financial data. Security-focused, local data processing, NO DATA LEAKED. Zero File Leakage AI Chatbot built with Privacy by Design. Documents are processed locally on your server—only extracted plain text (max 30KB) is sent to AI providers, never the original files. Supports PDF (with OCR), Word, Excel, images, and 20+ code formats. Multi-provider AI with DeepSeek, OpenAI GPT-4o, and Anthropic Claude. Features streaming responses via Vercel AI SDK 6, LaTeX/KaTeX math rendering, syntax highlighting, auto-continue for truncated responses, multi-session chat history, and real-time connection status. Built with Next.js 16, React 19, TypeScript strict mode, and Tailwind CSS 4.
Next.js 16
React 19
TypeScript
Vercel AI SDK
Tesseract OCR
Privacy by Design
MIT License
|
Building and training neural networks with PyTorch. Designing autonomous AI agents capable of multi-step reasoning, tool use, and self-correction. Multi-vector embeddings for image retrieval and semantic search using vector databases.
Applying large language models for content generation, summarization, and analysis. Implementing guardrails, input/output validation, and safety layers to ensure responsible and reliable AI deployment in production systems.
Crafting effective prompts for ChatGPT and other LLMs. Building AI-powered workflows and integrating language models into applications for automation, code generation, and intelligent data processing.
Leading 24/7 monitoring, incident response, and proactive maintenance for nationwide communication networks including VSAT satellite, terrestrial, and cellular infrastructures. Ensuring high availability and performance optimization for thousands of remote banking units across Indonesia.
Monitored and ensured 24/7 availability of over 20,000 ATMs, 200,000 EDCs, 500 CDMs, and CRM systems across the national banking network. Developed digital dashboards for real-time monitoring, root cause analysis, and operational reliability of electronic channel infrastructure.
Applied AES encryption for front-end code obfuscation to protect web applications. Deep expertise in cryptographic algorithms, secure coding practices, and vulnerability mitigation for banking-grade systems.
Deployed and operated firewall, IDS, IPS, anti-spam, and antivirus systems for banking environments. Built enterprise Active Directory infrastructure managing authentication and access control for large-scale organizations.
Designed multi-factor authentication systems using one-time passwords with MD5 hashing. Foundation in algorithm design, network protocols, and informatics engineering.
