PROJECTS

BY LUKA BUDIK

Advomate

Overview

Advomate is an AI-powered legal assistant that helps lawyers draft documents, research case law, and manage client information. It leverages advanced LLM technology to automate legal processes and improve efficiency.

Next.jsTypeScriptPythonOpenAIVector Databases

Challenge

Creating an AI system for legal terminology with confidentiality requirements. Legal documents are complex, highly specialized, and contain sensitive information.

Solution

Custom LLM fine-tuned on legal documents with secure data pipeline. Implemented advanced RAG system for accurate retrieval of legal information.

Results

70% reduction in document preparation time, 85% improved research accuracy. Over 300 clients including major companies like Škoda Auto and Skanska.

Key Features

  • AI-powered legal research with source citations
  • Document analysis and risk assessment
  • Semantic search across legal documents
  • Automated document generation
  • Secure handling of confidential information

AI Code Assistant

Overview

VSCode extension that provides intelligent code suggestions, refactoring, and documentation generation using LLMs. Enhances developer productivity by automating routine coding tasks and providing context-aware assistance.

TypeScriptVSCode APILangChainONNX Runtime

Challenge

Integrating LLMs with IDE workflows while maintaining performance and privacy. Balancing the need for powerful AI capabilities with local processing requirements.

Solution

Local model inference with context-aware code analysis and custom prompt engineering. Optimized for performance using ONNX runtime and efficient tokenization.

Results

50,000+ downloads, 4.8/5 rating, featured in VSCode marketplace. Significant productivity improvements reported by users across various programming languages.

Key Features

  • Intelligent code completion with context awareness
  • Automated code refactoring suggestions
  • Documentation generation for functions and classes
  • Code explanation and simplification
  • Support for multiple programming languages

SmartChat

Overview

Open-source chatbot framework with multi-modal capabilities, RAG integration, and fine-tuning tools. Provides a flexible foundation for building advanced conversational AI applications with minimal setup.

PythonFastAPIReactVector Databases

Challenge

Building a flexible framework that supports various LLMs and knowledge bases. Creating a system that can be easily extended and customized for different use cases.

Solution

Modular architecture with plugin system and adapter patterns for model/data source integration. Comprehensive API for extending functionality and integrating with existing systems.

Results

2,500+ GitHub stars, adopted by 15+ companies for production use cases. Active community contributing extensions and improvements to the core framework.

Key Features

  • Support for multiple LLM providers (OpenAI, Anthropic, local models)
  • Built-in RAG capabilities with various vector database integrations
  • Multi-modal support for text, images, and audio
  • Fine-tuning tools for customizing models
  • Comprehensive monitoring and analytics