Vaios Laschos

Vaios Laschos

Mathematician & Machine Learning Researcher

About Me

Mathematician with extensive postdoctoral experience, specializing in the intersection of mathematics and machine learning. Background in Pure Mathematics with a PhD in Applied Mathematics, focusing on Large Deviation Theory and Gradient Flows on spaces of probability measures (a.k.a Wasserstein space). Evolved from studying abstract mathematical questions to applied fields with connections to natural sciences and machine learning.

During my career I changed various research topics in an never ending quest of finding a topic that exhibits the golden ratio between challenge and impact. Currently I am working in a group where Bayesian tools are used for machine learning and general optimization. In the last two years, I have been polishing my ML skills with an emphasis on everything related to Large Language Models (LLMs), including developing various projects (portfolio available at Metaskepsis.com) and training LLMs for reasoning (portfolio available at huggingface.co/Metaskepsis). Currently running a seminar for my entire organization (WIAS) where I educate colleagues on the uses of LLMs with the scope of facilitating new research on the topic.

Academic Appointments

2021 - Present

WIAS

Postdoctoral Researcher

Bayesian methods on optimal transport, optimal transport and rough paths, algorithms on discrete optimal transport, EVIs on spaces of measures.

2018 - 2020

Technical University of Berlin

Postdoctoral Researcher

Risk sensitive decision making, optimal transport and machine learning, reinforcement learning and meta-learning techniques.

2015 - 2017

WIAS

Postdoctoral Researcher

Metrics on spaces of probability measures.

2013 - 2015

Brown University

Postdoctoral Researcher

Large deviations for Gibbs configurations, multi-agent risk sensitive stochastic control.

2013

MPI Leipzig

Guest Postdoctoral Researcher

Solutions of the Euler equation as quasigeodesics on the Wasserstein space.

Education

2009 - 2013

University of Bath

PhD in Applied Mathematics

Thesis: Wasserstein gradient flows via large deviations from thermodynamic limits of independent particle models.

2005 - 2009

Aristotle University of Thessaloniki

Master in Pure Mathematics

Dissertation: Potential theory and Brownian motion.

2000 - 2005

Aristotle University of Thessaloniki

Major in Pure Mathematics

Focus on Real Analysis.

Research

Publication Summary

12
Journal Publications
1
Conference Publication
2
Submitted Papers
2010-2024
Publication Years

Mathematical Foundations

  • Large Deviation Principle
  • Gradient Flows
  • Metrics on Spaces of Probabilities
  • PDEs
  • Fractal sets
  • Optimal Transportation

Machine Learning & AI

  • Large Language Models (LLMs)
  • LLM Training for Reasoning
  • Optimal Transport for Machine Learning
  • Generative Networks with Optimal Transport
  • Reinforcement Learning
  • Meta-learning
  • Generative Adversarial Networks
  • Few-Shot Learning
  • Bayesian Methods

Optimization & Control Theory

  • Risk-Sensitive Decision Making
  • Stochastic Control Theory
  • Multi-agent Systems
  • Optimal Transport Applications

Key Research Contributions

My research has advanced understanding in several areas:

  • Developed new geometric properties of cones with applications on the Hellinger-Kantorovich space
  • Established gradient flow structures for McKean-Vlasov equations
  • Advanced exit time risk-sensitive control for cooperative agents
  • Analyzed large deviations for configurations generated by Gibbs distributions

Teaching & Mentorship

Teaching Experience

Extensive teaching experience across multiple institutions, including tutorials in Calculus and Linear Algebra as a PhD student. Served as coordinator of the Probability seminar at Brown University for two consecutive years. At TU-Berlin, conducted several seminars on advanced topics in Reinforcement Learning and co-organized a course where students completed projects related to Machine Learning or Computational Neuroscience.

Student Supervision

Supervised 20+ Master's theses (2019-2024) across diverse topics in machine learning, reinforcement learning, and optimal transport applications.

Machine Learning & AI

  • LLM performance optimization
  • Prompt engineering with RL
  • Representation learning
  • Few-shot classification

Reinforcement Learning

  • Generative algorithms
  • Deep neuroevolution
  • Ad-hoc cooperation
  • Risk-sensitive RL

Optimal Transport

  • Generative adversarial learning
  • Entropic optimal transport
  • Sinkhorn algorithm initializations

Skills

Technical Skills

Operating Systems 5.5/6
LaTeX 5.5/6
Git 4.5/6
Python 4/6
Haskell 3.5/6

Languages

Greek 5.5/6
English 5/6
German 3/6
Spanish 3/6

AI & Machine Learning

Train NN of various architectures 5/6
Prompt optimization 5/6
Create projects with vibe-coding 5/6
PyTorch 4.5/6
Fine Tune LLMs 4/6

Soft Skills

Listening 6/6
Problem Solving 5.5/6
Teamwork 5.5/6
Flexibility 5/6
Public Speaking 2/6

Personal Interests

Climbing Yoga Travelling Cooking Trying random sports Making decisions that take me far away from my comfort zone Complaining about those decisions