Martin Studna

• Develop computer vision algorithms for analyzing eSports videos to enhance game analysis for Data Science team
• Design algorithms for detecting ball possession, player actions, and key in-game events
• Utilize Python, PyTorch, and TensorFlow as primary deep learning frameworks for ongoing algorithm development and refinement

I designed an AI agent that processes client requirements and generates a design brief for designers. The agent evaluates the completeness of the request, providing a score from 0% to 100% based on the available information. If any details are missing, it notifies the client and suggests improvements.

I built the solution using Python, LangGraph, and LangSmith, leveraging SambaNova for serverless inference. The entire system is deployed on AWS for scalability and reliability.

Automated Stone Slab Matching System

For Stonify - US based company, I designed an AI-driven system that automates stone slab selection. This project was delivered with collaboration with a company Cod8. Using Qdrant vector search, the AI agent analyzes a client’s slab image and retrieves the five most similar options, generating an instant offer. This eliminated manual work, making the process faster and more efficient.

• Designing and implementing LLM solution for document compliance
• Evaluating and fine-tuning LLMs
• Designing Graph RAG (Neo4j) database
• Experimenting with open-source models such as Llama and Mistral
• Deploying open-source LLMs with Triton Inference Server

• Actively engaged in the development of advanced AI tools for a sophisticated camera surveillance system
• Key responsibilities include the design and implementation of models for action recognition, pose estimation, person identification, and tracking
• Utilize a diverse toolkit comprising Python, TensorFlow, PyTorch for model development, supplemented by TensorRT and C++ for writing high-performance inference engines

• Led the design and implementation of a comprehensive system that integrated various cryptocurrency markets and exchanges, facilitating efficient crypto exchanges with minimal fees via a mobile application
• Utilized a robust technology stack including C#, Microsoft SQL, Cosmos DB, Redis, SendGrid, and NService Bus, ensuring the system's high performance and security standards
• Played a crucial role in developing and deploying the system on the Azure cloud platform, overseeing the project's DevOps aspects

• Implemented studiooo.cz website, aimed at providing educational workshops on IT, internet security, and AI awareness
• Leveraged JavaScript and Vue for crafting a user-friendly frontend, effectively communicating workshop details to users
• Built a robust backend system using NodeJs and NestJs, coupled with PostgreSQL for an efficient database solution
• Focused on creating a CRUD application for streamlined workshop management, aligning with the company's mission to educate the public about technology's societal impact

• Specialized in designing and implementing chatbot solutions for various clients
• Utilized Vue.js for frontend development and Node.js for backend integration with IBM Watson API
• Key projects included developing the Vodafone Tobi chatbot and Anežka, a virtual assistant for COVID-19 and emergency information in Czechia
• Gained significant experience in chatbot development, honing skills in Vue.js and Node.js, and contributing to innovative customer service solutions

• Tasked with developing a cloud-based system for smart house devices
• Employed skills in C#, ASP.NET, and MongoDB to contribute to both the design and implementation of the system