Hai there! :-) I'm Libor Burian or Burny! I'm looking to collaborate! :-) ## Focus - [[Artificial Intelligence|ML and AI research and engineering ]] - Technical writing - Community building ![[Images/dd8ede9a226ac6782d4f102f2e25224a_MD5.jpeg|200]]![[Images/35727f7355f3216747dce9cc72ebe34b_MD5.jpg|200]] ## I am constantly following, reading, writing about, playing with, expanding, etc. on various new research papers and industry news, with others, in: - [[Artificial Intelligence|AI progress, such as new LLMs, novel AI architectures, agents, reinforcement learning, etc.]] - [[Mathematical theory of artificial intelligence|The science of AI/ML/deep learning]], [[Mechanistic interpretability|AI interpretability]] that tries to reverse engineer how AI systems work, etc., - [[AI engineering]], applying AI in industry, to [[Artificial intelligence x Programming Coding Software Engineering|software engineering]], to [[Artificial Intelligence x Physics|physics]], to other [[Artificial intelligence x Engineering|engineering]], to other [[Artificial intelligence x Science|sciences]], etc. - [[Artificial Intelligence|And other types of AI research papers and industry news]] - I'm highly curious and also like to sometimes explore some [[Mathematics|math]], [[physics]], science of [[intelligence]], [[cognitive science]], [[neuroscience]], [[philosophy]], [[computer science]], [[Future of humanity, AI, sentience, futurology, politics|futurology]], scifi, etc. - See my [[projects]], my [AI interpretability wiki](https://aiinterpretability.miraheze.org/wiki/Main_Page), my [[Home|exocortex]] and its [[artificial intelligence]], [[mathematical theory of artificial intelligence]], [[AI engineering]], and other pages - I consider myself a fast, adaptable learner! When I'm committed to something, I become fully immersed in it, obsessed by it, it becomes my reality, and I give it my full effort! ## Experience - For 3+ years following various new AI research papers and industry news - For 2+ years working in AI industry on LLMs, agents, etc. for various clients - For 10+ years all sorts of community building activities - For 10+ years general education and experimenting and curiosity about [[STEM]] related things, mostly computer science ## Education - Since 2020: [CTU FIT](https://en.wikipedia.org/wiki/Faculty_of_Information_Technology,_Czech_Technical_University_in_Prague) computer science university, took some courses also from [MFF CUNI](https://www.mff.cuni.cz/en) and [CTU FEE](https://fel.cvut.cz/en). 4 years of IT high school before that. - [[Artificial Intelligence|Going through more online courses from MIT, Stanford, and other universities and other resources]] ## Community building I do and did - Almost 20 000 followers on X (Twitter): [Burny - Effective Curiosity](https://x.com/burny_tech) with a community, used to operate Youtube channel with 35000 subscribers with a community, and livestreamed. - Moderating bunch of active discord servers - [Mechanistic Interpretability](https://mechinterp.com/) with [weekly research paper reading group](https://www.youtube.com/@MIDiscordServerRecordings), paper presentations by authors of papers, and other things - [Machine Learning Street Talk](https://www.youtube.com/c/MachineLearningStreetTalk) with weekly debates I coorganize and other things. Also helped with bunch of their videos on their YouTube channel - Created bunch of discord servers that are/were active - [Qualia Research Institute](https://qri.org/) where I cohosted community events with main researcher - Burny's server where I interacted with my own community) - and moderated more in the past - Active in other communities on Discord and other social media: AI research communities (such as [EleutherAI](https://en.wikipedia.org/wiki/EleutherAI), [Yannic Kilcher](https://www.youtube.com/yannickilcher), [Latent Space](https://www.youtube.com/@LatentSpacePod)) and other communities offline and online ([Effective Altruism](https://www.effectivealtruism.org/) (where I gave a presentation, and attended EAGx a few times, and other events), [LessWrong](https://www.lesswrong.com/)(attended LessWrong Community Weekend), [AstralCodexTen](https://www.astralcodexten.com/) (attended meetups), tpot,…), Czech universities I just mentioned, and some other nerdy and geeky communities, where I helped with some community events (I always love connecting compatible people together!) - Helped [Czech Society for Cognitive Science](https://kognitivniveda.cz/) with some events with researchers presenting their work and other things ## Wikis I work on - [AI interpretability wiki](https://aiinterpretability.miraheze.org/wiki/Main_Page): Wiki of the AI interpretability scientific field. AI interpretability tries to understand how AI systems work using science. These findings help causally predict and explain the behavior and evolution of AI systems, making them more interpretable, transparent, steerable, safe, and so on. It does so by empirically finding representations and circuits, as in biology or reverse engineering of software (the mechanistic interpretability subfield); by empirically finding mathematical laws, as in physics, or by analytically deriving mathematical properties, as in theoretical physics (the deep learning theory subfield). - [[Exocortex]]: My personal wiki of everything I am curious about that I explore. Full of interconnected topics, taxonomies, links to resources, written down thoughts, ideas, articles, projects, and so on. ## Some of the projects I work/worked on #### Research primarily for science ##### Artificial intelligence / machine learning ###### Mechanistic Interpretability - [Attention Head Zoo: 2-Layer Attention-Only Transformer](https://github.com/BurnyCoder/attention-head-zoo-2-layer-attention-only-transformer): Manually cataloguing and classifying the functional roles of all 24 attention heads in a 2-layer attention-only transformer, using TransformerLens and circuitsvis for mechanistic interpretability. - [LLM steering vectors for physics](https://github.com/BurnyCoder/llm-steering-vectors-for-physics): Finding and applying steering vectors to LLMs to increase performance on physics problems. LLM steering vectors are directions in a model's activation space that, when added to its hidden states at inference time, push the model's behavior toward or away from a specific concept or trait (like honesty, refusal, or sycophancy) without retraining the weights. - Projects at very start - Mechanistic interpretability of text to speech models using sparse autoencoders: Sparse autoencoders learn sets of sparsely activating features that are more interpretable and monosemantic than directions identified by alternative mechanistic interpretability approaches. - Mechanistic interpretability of [physics foundation models](https://arxiv.org/abs/2509.13805) using steering vectors and sparse autoencoders - Reverse engineering a toy transformer grokking on a modular multiplication task based on [modular addition](https://arxiv.org/abs/2301.05217) - Mechanistic interpretability in time as LLMs are trained: How do circuits emerge - Mechanistic interpretability on [mHC: Manifold-Constrained Hyper-Connections](https://arxiv.org/abs/2512.24880) - Autoresearch on various mechanistic interpretability projects - Various steering vector experiments ###### Other - [Autoresearch Bootstrap](https://github.com/BurnyCoder/autoresearch-bootstrap): An autonomous LLM agent for running AI research end to end with minimal human intervention. It works by an agent such as Codex or Claude Code looping over specs specifying AI researcher workflow. Expansion of Andrej Karpathy's autoresearch setup. - [Auto Picbreeder](https://github.com/BurnyCoder/auto-picbreeder): Picbreeder but LLMs can play the role of humans. Picbreeder evolves images without any training data using Compositional Pattern Producing Networks (CPPNs) evolved by NEAT (NeuroEvolution of Augmenting Topologies). - [Diverse Group Relative Policy Optimization (DGRPO)](https://github.com/BurnyCoder/diverse-group-relative-policy-optimization)): Reinforcement learning algorithm to make LLMs reason more creatively by incorporating solution diversity into the LLM RL GRPO advantage calculation through upweighting less likely but correct solutions to incentivize rare solutions - [LLMs take autism test](https://github.com/BurnyCoder/llms-take-autism-test): Benchmark in which various current LLMs answer sampled items from the Autism Spectrum Quotient through OpenRouter, then aggregates the results into reproducible artifacts and a LaTeX paper. - [Automated scientific method using large language models](https://github.com/BurnyCoder/llm-scientific-method) - Projects at very start - Adding transformer to Hamiltonian neural networks - Merging intelligence definitions #### Engineering primarily for science - Projects at very start - Top liked [alphaxiv](https://www.alphaxiv.org/) papers clustered #### Research primarily for industry - AI Agent company: A hierarchical team of AI agents that form a company with CEO, CTO, engineers, and so on. CEO recruits new agents as the company runs. Orchestrator adds requested MCP tools. Powered by Claude Code agents, Codex agents, OpenClaw agents, Hermes agents, or OpenCode agents. - [Multi LLM synthesis](https://github.com/BurnyCoder/llm-synthesis): An AI system that synthesizes responses from multiple Large Language Models (LLMs) to provide more comprehensive and reliable answers. #### Engineering primarily for industry - [LLM RL - Language Model Reinforcement Learning Pipeline](https://github.com/BurnyCoder/large-language-model-reinforcement-learning-pipeline): Training language models using reinforcement learning covering SFT, Reward Modeling, DPO, and GRPO built on TRL (Transformer Reinforcement Learning) from Hugging Face. - AI Agent GUI specialized for company's usecase: Research company's products, social media content planning, competitor analysis, customer-response drafting, and internal knowledge base building. - Customer support phone AI agent for planet ticket company, plus simulating employee agent and customer agent interacting - AI Agent explorer: Platform to explore and interact with multiple AI agents including a MultiAgent AI Coding system (for collaborative code generation and review), a Deep Research agent (for thorough research with clarifying questions), and a Synthesis agent (for combining responses from multiple LLMs) with a modern React frontend and Python backend. - [Software for autonomous drones](https://github.com/BurnyCoder/DJI-MSDK-to-PC) Autonomous drone using AI agent with vision analysis capabilities, realtime person tracking, live control using keyboard, live video stream, live queries data. - [Syri](https://github.com/BurnyCoder/syri): AI voice assistant that uses web browser use with Claude via Portkey, AssemblyAI for speech-to-text and ElevenLabs for text-to-speech - [Large language model evaluations](https://github.com/BurnyCoder/llm-evals): Evaluating Large Language Models. Generate test questions on a given topic. Provide your own custom questions. Customize the evaluation prompt. View evaluation results and average scores. - Brand statistics: A real-time brand analytics dashboard built with Reflex that tracks brand sentiment, visibility, and share of voice across AI models using OpenAI GPT-5 with web search capabilities. - Selfcorrecting multi agent threat hunting system - OCR/LLM based invoice processing #### Engineering primarily for myself ##### Artificial intelligence / machine learning - [Jarvis Waifu Supervisor](https://github.com/BurnyCoder/jarvis-waifu-supervisor): AI-powered productivity monitor that watches your screen and webcam to keep you focused on your tasks. Deep work loop with Vision LLM and TTS, blocking websites and apps. - [Wikipedia AI agent research assistant](https://github.com/BurnyCoder/wikipedia-ai-agent) Wikipedia AI agent research assistant. LangChain's LangGraph's ReAct agent architecture, LLMs (OpenAI, Anthropic, Google), Wikipedia API, RAG with FAISS vector db, semantic chunking, GraphRAG, Streamlit frontend, terminal and web interfaces - [Speech to Text (OpenAI gpt-4o-transcribe)](https://github.com/BurnyCoder/speech-to-text-openai): Transcribe audio files using OpenAI's GPT-4o-transcribe model using OpenAI API. - [FL Studio Piano Roll MCP](https://github.com/BurnyCoder/fl-studio-mcp): Fork of calvinw/fl-studio-mcp with improved Windows support (FL64.exe detection, window class name fix). An MCP (Model Context Protocol) server that enables AI assistants like Claude to interact with FL Studio's piano roll. Create melodies, chord progressions, and musical patterns through natural language conversation with automatic, real-time updates. - [AI Youtube Assistant](https://github.com/BurnyCoder/ai-youtube-assistant): AI assitant that let's you talk about the contents of Youtube videos using Langchain, Streamlit, FAISS, GPT3.5 ##### Other - [Multi-Interval Habit Tracker](https://github.com/BurnyCoder/multi-interval-habit-tracker): Additional functionality to complement Habitica, a browser-based app for tracking habits that are completed on a recurring schedule (e.g. every N days), persisting data in your project folder via a local backend service. - [Python word cloud SVG generator](https://github.com/BurnyCoder/word-cloud) - [Wikipedia Knowledge Graph Visualization](https://github.com/BurnyCoder/wikipedia-knowledge-graph): Interactive visualization of Wikipedia article connections as a knowledge graph - [Interactive A* (A-star) pathfinding simulation](https://github.com/BurnyCoder/interactive-a-star-path-finding-visualization) - [Nested Skill Tree](https://github.com/BurnyCoder/nested-skill-tree): A simple application that displays a hierarchical skill tree with the ability to mark skills as completed and expand/collapse subtrees. Built with Python's standard tkinter library. - Random fun with physics simulations, artificial life, and other math: fluid dynamics, solar system simulation, celluar automata, diffusion, bird flocking - Random fun with game development: [Flying Tank Glitch Sandbox](https://github.com/BurnyCoder/flying-tank-glitch-sandbox), [roguelike dungeon crawler game with procedurally generated world and LLM NPCs](https://github.com/BurnyCoder/llm-dungeon-roguelike), [bunnyhopping](https://github.com/BurnyCoder/bunnyhopping), [racing](https://github.com/BurnyCoder/racing-game), replicating Minecraft, GTA, terraria, tower defense, and other games #### Learning ##### Artificial intelligence / machine learning - [AI From Scratch](https://github.com/BurnyCoder/ai-from-scratch): This repository contains implementations of various AI and machine learning concepts, architectures, and exercises built from scratch. It serves as a learning resource for understanding the underlying principles of artificial intelligence and machine learning algorithms. Transformers, LLM, GPT-2, LSTM, Diffusion, U-Net, CNNs, MLP, Reinforcement learning, Bigram, Linear+Polynomial+Logistic Regression, GRPO, Selfplay, MCTS, PyTorch, Scikit-learn, Torchvision, NumPy, Matplotlib, Einops, Transformers, Plotly - [Practical AI Projects](https://github.com/BurnyCoder/practical-ai-projects): Collection of practical AI applications and implementations using various machine learning, deep learning, and natural language processing techniques. LLM (training, finetuning, reasoning reinforcement learning , RAG, multiagents), image classification&segmentation, text&image answering, text2speech, movie recommendation, dimensionality reduction, LlamaIndex, Autogen, PyTorch, TensorFlow, Transformers, TRL, Keras, fastai, NumPy, Skicit-learn, OpenAI, ElevenLabs, ResNet, LSTM, Autoencoder, SVM - [Interactive bayesian inference visualization](https://github.com/BurnyCoder/interactive-bayesian-inference-visualization): Interactive visualization of how the probable delivery date of a baby changes over time using Bayesian inference, updating the distribution based on the current day of gestation. Based on problem from Stanford CS109. - A neural network that learns to play Snake game using Deep Q-Learning ##### Physics - [Electromagnetism vector field curl visualization](https://github.com/BurnyCoder/electromagnetism-vector-field-curl-visualization): Interactive 3D visualization of electromagnetic gauge freedom. Demonstrates how different vector potentials (Landau gauges A₁ and A₂) produce the same magnetic field B, with toggleable field layers, adjustable parameters, and orbital camera controls. Using Three.js and React. - [Angular momentum visualization](https://github.com/BurnyCoder/angular-momentum-visualization): Interactive 3D visualization of angular momentum that shows a particle in motion with its position, momentum, and angular momentum vectors. - [Pendulum phase space flow](https://github.com/BurnyCoder/pendulum-phase-space-flow): Phase space of pendulum (angle vs. angular velocity) with multiple trajectories to illustrate the overall flow. - [Hamiltonian harmonic oscillator](https://github.com/BurnyCoder/hamiltonian-harmonic-oscillator): Harmonic oscillator simulation based on the Hamiltonian mechanics equations, visualizing the oscillating mass and displaying the key equations of motion, real-time values for position, momentum, and total energy, with sliders to adjust the mass and the spring stiffness to see how they affect the oscillation. - [Two body coupled oscillation](https://github.com/BurnyCoder/two-body-coupled-oscillation) : Interactive simulation of two masses connected by a spring, demonstrating the principles of coupled oscillation with user-adjustable mass, spring stiffness, and damping. - [Visalizing energy langrangian hamiltonian action](https://github.com/BurnyCoder/visalizing-energy-langrangian-hamiltonian-action): Explore kinetic energy, potential energy, lagrangian, hamiltonian, action with real-time graphs and equations. ##### Web development - [Amazon clone using MERN stack](https://github.com/BurnyCoder/amazon-clone): React Javascript library/framework with Redux, Material UI, React Router Dom, React Currency Format, Firebase, Stripe, Axios, CSS, HTML See more at [[projects]] and [BurnyCoder (Burny) · GitHub](https://github.com/BurnyCoder) ## Questions I love to explore - How does [[artificial intelligence]] [[Mechanistic interpretability|work]]? What's the current state of empirical research and [[Mathematical theory of artificial intelligence|mathematical theory]] in [[artificial intelligence]]? What is the state of the art in [[artificial intelligence]] [[AI engineering|engineering practice]]? - How to define and understand [[artificial general intelligence]] and [[superintelligence]]? How to [[AI safety|make it do what we want]], how to steer it, how to align it? - How to [[Artificial Intelligence#Crossovers Omnidisciplionarity|apply AI]] for reverse engineering [[Applied mathematics|mathematics behind everything]]? How to [[Artificial Intelligence#Crossovers Omnidisciplionarity|apply AI]] for good as ideally as possible as much as possible? - What is the fundamental [[mathematics]] of [[intelligence]]? What are all the different types of all the possible current and future intelligent systems? - What is the fundamental [[mathematics]] of the [[Cognitive science|brain]]? How to upgrade [[Human intelligence amplification|human intelligence]]? - How does [[Artificial Intelligence x Biological Intelligence|AI and biological intelligence compare?]] How can [[Biological intelligence|humans]] and [[Artificial Intelligence|AIs]] form even greater [[collective intelligence]]? - How does the [[Natural science|world]] work? How does [[everything]] [[Omnidisciplinarity|work]]? - What is the [[Theory of Everything|fundamental mathematics]] of the [[Physics|universe]]? What are all the equations, and [[Mathematics|mathematical]] structures more generally, governing [[Science|reality across all scales]] in physics, and in all [[Natural science|natural science]] more generally? - How to connect all [[Science|sciences]], [[Formal science|formal]] and [[Natural science|natural]]? What is the fundamental [[mathematics]] behind [[emergence]] and [[complexity]]? How does [[biology]] and other scientific fields emerge from [[physics]] and [[chemistry]]? - How to do good [[science]] using [[philosophy of science]], [[rationalism]], and other bodies of knowledge? - What is the fundamental [[mathematics]] of [[creativity]] in [[science]] and [[art]]? How to make machines creative beyond human limitations and comprehension for [[Science|scientific]] discovery, [[Artificial Intelligence x Physics|physics]], [[mathematics]], [[art]], [[philosophy]]? - What are all the concepts in [[mathematics]]? What are all the possible [[Foundations of mathematics|foundations]] and [[mathematics]] with all sorts of mathematical universes and which ones are the best in what contexts? - What is the fundamental [[mathematics]] of [[Future of humanity, AI, sentience, futurology, politics|building a great future for all where everyone flourishes]]? How to make the world better for all? How to maximize the benefits, and minimize the disadvantages, of [[Technology|technologies]] and [[Politics|political systems]]? How to think about philosophical movements such as [[effective altruism]] and [[effective accelerationism]]? What is and what will be the geopolitics of AI? What are the probabilities of different future scenarios? - What is the fundamental [[mathematics]] of [[consciousness]]? How to work with mind and find truth in scientific secular way? - What are the answers to the problems in [[philosophy]]? - See more at [[exocortex]]. Also see my [YouTube channel](https://www.youtube.com/@burnytech). I also like cute things! ![[Images/52c0b8e6fcb51e527f12db5af187083d_MD5.svg]] ## Contact me - See [[Contact and links|contact and links]] ### Rest of this website - [[About me and my experience]] - [[Exocortex]] - [[Exocortex/Unsorted/Projects|Projects]] - [[Contact and links]] - [[Videos]]