AI/ML Projects

ComfyUI Mastery

Extensive experience developing complex workflows within ComfyUI, utilizing a wide array of models:

• Image Generation: Expert implementation of cutting-edge models including SDXL, Flux, and Hi-Dream for high-quality image creation.
• Video Generation: Proficiency with advanced video models such as Wan 2.1, Hunyuan, and Cog Video for dynamic content creation.
• Specialized Techniques: Implementation of Reactor nodes for efficient face replacement and various upscale models to significantly enhance image detail and resolution.
• LORA Training: Skilled in training Low-Rank Adaptation (LORA) models for Flux, utilizing both ComfyUI and Fluxgym platforms to create specialized model adaptations.

Dataset Preparation and Captioning Tool

Developed a custom software solution for preparing image datasets and generating captions. This tool leverages local AI models such as Llama 3.2 Vision and Google Gemini Flash 2.0 for efficient and localized processing without relying on external APIs.

Advanced implementation of automation workflows and integrations:

• Advanced RAG Pipelines: Created sophisticated workflows for Retrieval Augmented Generation (RAG) using n8n, enabling more contextually aware and accurate AI responses.
• Workflow Automation: Developed various automation solutions to streamline processes and enhance productivity.
• Website Chatbots: Built custom website chatbots, including personal implementation, powered by advanced RAG and AI agent pipelines orchestrated within n8n.

Strong foundation in Python, while effectively utilizing AI code generation tools like Cursor and the VS Code extension Cline to develop software and tools in various programming languages:

Actively using several tools for continuous learning, experimentation, and self-improvement in the AI space:

• NotebookLM: Utilized for synthesizing information from various documents, research papers, and articles to deepen understanding of complex AI/ML concepts.
• Ollama: Used for running open-source large language models (LLMs) locally, enabling experimentation with different models and developing applications that require offline language processing.
• LM Studio: Provides a user-friendly interface for discovering, downloading, and running a wide range of LLMs on a local machine, facilitating rapid prototyping and offline inference.

• MCP Servers and AI Agents: Familiar with the concepts and efficient utilization of MCP servers and AI agents for scaling workflows and deploying AI applications.
• Prompt Engineering: Highly skilled in crafting effective prompts to guide AI models and tools, consistently achieving desired outputs with reduced iteration and token usage.