I’m Stanislav Chekmenev, a machine learning specialist & educator. My professional path led me through particle physics research to cutting-edge ML applications, with a current focus on symmetry-aware models and generative ML for protein design. I hold a PhD in Physics from RWTH Aachen University, where I specialized in spin dynamics simulations of subatomic particles. With this solid background in math and physics, I smoothly transitioned to machine learning after my defense.
Professional Evolution
Over the past eight years, I’ve evolved from traditional data science roles into advanced ML positions, consistently studying the latest techniques and implementing them at work.
Teaching & Mentoring
Currently working as an Instructor & Mentor at Data Science Retreat, where I teach courses on geometric deep learning and supervise portfolio projects mainly connected to GNNs and biomedical applications. Seeing students understand and apply complex ML concepts is incredibly rewarding to me.
Engineering Industry Projects
As a Founding ML Scientist at Qypt, I led development of a privacy-preserving LLM-based assistant, resulting in the first released version of the app.
Cutting-edge ML Research
- Certainty Lab: Co-developed a parallelized variant of MuZero and wrote a scalable GNN architecture for real-world RL traffic control, achieving 10x increase in sample efficiency.
- Hella Aglaia: Created a proof-of-concept of a real-time anomaly detection system for quality control using self-supervised learning and normalizing flows. This POC evolved into a fully working product.
Beginning of Data Science Career
I built predictive models for sales forcasting and customer behavior analysis at two startups: HeilpflanzenWohl and Looping PowerRoom.
Academic Years
Completed PhD in Physics at RWTH Aachen, focusing on spin dynamics and mathematical modelling.
Current Focus
I’m currently focused on geometric deep learning and protein design, which I began exploring in 2023. At the end of 2024, I took a break from full-time work to concentrate on the self-initiated educational research project in protein generation. I’ve successfully studied and modified one of the recent SOTA models in the field, FoldFlow-2.
Philosophy & Approach
Apart from my technical interests, I share a deep appreciation for Eastern philosophy and spirituality. After spending almost a year at a Buddhist monastery in 2023, I developed a new perspective on patience, kindness and the value of sustained focus. I learned how to translate these qualities not only into my day-to-day life, but especially into my profession. Combining this experience with my physics background, I now tackle problems both patiently and foundationally, by searching for an effective solution rather than a quick fix.
Additionally, I’ve guided numerous students and colleagues through complex ML projects, always emphasizing clear understanding of the core principles and the importance of possible ethical implications.
Selected Publications in Spin Dynamics
- Estimation of Systematic Errors for Deuteron Electric Dipole Moment Search at COSY, International Journal of Modern Physics: Conference Series Vol. 40, 2016.
- Quasi-frozen Spin Method for EDM Deuteron Search, IPAC, 2015.
- Investigation of Lattice for Deuteron EDM Ring, ICAP, 2015.
- Estimation of Systematic Errors for Deuteron Electric Dipole Moment (EDM) Search at a Storage Ring, IPAC, 2014.
Curriculum Vitae
For a comprehensive overview of my background, experience, and qualifications you can download my CV below.
Contacts
In case you’re interested in Bio ML, want to collaborate on a project or have some ML related questions, you can reach out to me via the following platforms.