Skip to content
View michailmitsakis's full-sized avatar
16 followers · 71 following

Block or report michailmitsakis

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please donโ€™t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this userโ€™s behavior. Learn more about reporting abuse.

Report abuse
michailmitsakis/README.md

Hi there ๐Ÿ‘‹, I'm Michalis

๐Ÿ‘จโ€๐Ÿ’ป About Me

I'm a motivated engineer and business consultant in the energy sector, experienced in and passionate about decarbonization, sustainable energy solutions and accelerating R&D for low-carbon technologies. Specialized in electrochemistry, clean hydrogen and machine learning for materials design. I am goal-driven, action-oriented and excel in a collaborative environment. Seeking to grow my skillset and contribute to the green transition through my generalist mindset, while finding creative solutions to complex problems.

๐Ÿ›  Tech Stack

Languages: Python

Libraries: Numpy, Pandas, Scikit-Learn, fastai, PyTorch, JAX/Flax

Bayesian Optimization: Dragonfly, Ax, BayBE, Honegumi, GPax, Pyro

AI/LLMs: Context & harness engineering. Experience with local workflows (RAG using Ollama/LMStudio/AnythingLLM and OpenWebUI) and associated custom MCPs & plugins.

Databases: Qdrant, PostgreSQL

IDEs: Cursor, VSCode

Other: Docker, Git, PowerBI, MATLAB

๐Ÿค Connect

Reach out to me on LinkedIn or my E-mail!

Pinned Loading

  1. Ax_bayes_opt_NiW Ax_bayes_opt_NiW Public

    Multi-objective Bayesian optimization for electrochemical experiments with Ax.

    Jupyter Notebook

  2. baybe_project-surface-science-syndicate baybe_project-surface-science-syndicate Public

    Project from Team 7 - Surface Science Syndicate; developed in competition for the 2024 Acceleration Consortium Bayesian Optimization Hackathon. Achieved 3rd place out of 44 competitors.

    Jupyter Notebook

  3. CaMEL-RAG CaMEL-RAG Public

    Code4Catalysis project developed in competition for the 2025 LLM Hackathon for Materials and Chemistry.

    Jupyter Notebook

  4. fantasy-book-assistant fantasy-book-assistant Public

    This is your friendly assistant for helping you pick your favorite fantasy and sci-fi books. It's a RAG app built built as part of the LLM Zoomcamp 2025 Edition.

    Jupyter Notebook

  5. notion-second-brain notion-second-brain Public

    A local, memory-aware, Second Brain agent. Ingests Notion pages, processes PDFs/images into markdown, indexes everything into Qdrant, and serves RAG queries, with all inference running locally thrโ€ฆ

    Python

  6. agentic_AI_MOOC_UC_Berkeley_2025.md agentic_AI_MOOC_UC_Berkeley_2025.md
    1 
    I'm excited to share the insights I gained from my favorite lecture in the **Agentic AI** (*CS294-196 Fall 2025*) course ([https://agenticai-learning.org/f25](https://agenticai-learning.org/f25)) by [UC Berkeleyโ€™s RDI Center on Decentralization & AI](https://rdi.berkeley.edu/): โ€œAI Agents to Automate Scientific Discoveriesโ€ by James Zou.
    2 
    
              
    
              
    
                3
                
    โ— First, it introduced the concept of using AI agents as โ€œ๐—ฐ๐—ผ-๐˜€๐—ฐ๐—ถ๐—ฒ๐—ป๐˜๐—ถ๐˜€๐˜๐˜€โ€ to automate and accelerate scientific discoveries. Modern AI agents are able to access specialized tools, databases and memory, enabling them to tackle versatile research problems, such as scientific hypothesis generation, experiment design, data analysis and paper writing.
    
              
    
              
    
                4
                
    
              
    
              
    
                5
                
    โ— Illustrating this capability is ๐—ง๐—ต๐—ฒ ๐—ฉ๐—ถ๐—ฟ๐˜๐˜‚๐—ฎ๐—น ๐—Ÿ๐—ฎ๐—ฏ, an AI-human collaboration for research. It mirrors a real research lab, with a โ€œprincipal investigatorโ€ agent overseeing specialized sub-agents with different โ€œbackgroundsโ€, while a human researcher provides high-level feedback. These agents are able to hold efficient group meetings with each other based on an agenda provided by the researcher, as well as individual meetings, where the researcher interacts with a single LLM to solve a particular task. Teams of multiple agents debate, leading to more creative and robust reasoning compared to a single agent working alone. 
    
              
    
          
    
    
    
  
    

    7.