Curriculum Vitae

Research Assistant

Caglar Gulcehre Lab for AI Research @ EPFL

Working on foundation models for decision-making problems, specifically on efficient RL for long-horizon problems with algorithm distillation

• Large Language Models
• State-Space Models
• Reinforcement Learning

Research Intern

InstaDeep

Pre-training Transformer models on a large offline reinforcement learning dataset (1+ billion transitions) of the classic Atari benchmark

• Jax
• Transformers
• Variational Auto-Encoders
• Offline Reinforcement Learning
• HDF5
• Google Cloud Platform

Research Assistant (Unofficial)

Laboratory for Information and Inference Systems @ EPFL

Co-authored a paper about self-supervised learning for combinatorial optimization problems and worked on hospital staff scheduling using deep reinforcement learning and graph neural networks

• Python
• Reinforcement Learning
• Graph Neural Networks
• Combinatorial Optimization
• Scheduling

AI Research Intern

Aucos AG

Developed solutions for production line optimization and multi-camera tracking

• Python
• Computer Vision
• Planning
• Graph Representational Learning
• Multi-agent Reinforcement Learning

MSc

École Polytechnique Fédérale de Lausanne (EPFL)

Data Science

• Visual Intelligence
• Reinforcement Learning
• Network Machine Learning
• Large-Scale Data Science For Real-World Data
• Statistics for Data Science
• Mathematics of Data: from Theory to Computation
• Intelligent Agents
• Machine Learning

BSc

Maastricht University

Data Science and Artificial Intelligence

Winner - Mini-Hackathon on Multimodal Apps and Training LLMs

LauzHack Association

Won the Organizer's prize for Tralala, a tool that enables users to generate and refine 3D models dynamically using natural inputs like text prompts, spoken descriptions, or reference images

Winner - LauzHack 2025

LauzHack Association

Second place in LauzHack 2025. Developed an on-device Flask-based web application for creating, managing, and visualizing notes effectively and securely using RAG and LLM models

Winner - HackUPC 2024

Polytechnic University of Catalonia

Won [MLH] Best Use of Taipy and Intersystems Challenge - Best use of GenAI using InterSystems IRIS Vector Search for WALL-M, an AI-powered email assistant

Research Scholar Assistant

EPFL

Highly competitive research program at the Caglar Gulcerhe's Lab for AI Research (CLAIRE)

Master's Excellence Fellowship, EPFL

EPFL

Two-year fellowship valued at 40,000 CHF, awarded to ~3% of EPFL master students based on outstanding academic records

CS-503 Visual Intelligence Best Project Award

EPFL

Secured first place among 14 teams for project 'Hunting for Insights: Investigating Predator-Prey Dynamics through Simulated Vision and Reinforcement Learning'

Best Bachelor's Thesis Award 2023

Maastricht University

University-wide award valued at 500 EUR for the best bachelor's thesis research among all other students in the programme

Algorithm Discovery With LLMs: Evolutionary Search Meets Reinforcement Learning

arXiv preprint

Discovering efficient algorithms for solving complex problems has been an outstanding challenge in mathematics and computer science, requiring substantial human expertise over the years. Recent advancements in evolutionary search with large language models (LLMs) have shown promise in accelerating the discovery of algorithms across various domains, particularly in mathematics and optimization.

Optimizing Job Allocation using Reinforcement Learning with Graph Neural Networks

arXiv preprint

Efficient job allocation in complex scheduling problems poses significant challenges in real-world applications. In this report, we propose a novel approach that leverages the power of Reinforcement Learning (RL) and Graph Neural Networks (GNNs) to tackle the Job Allocation Problem (JAP).

PASTA: Pretrained Action-State Transformer Agents

Reinforcement Learning Journal

Self-supervised learning has brought about a revolutionary paradigm shift in various computing domains, including NLP, vision, and biology. Recent approaches involve pre-training transformer models on vast amounts of unlabeled data, serving as a starting point for efficiently solving downstream tasks.

Maximum Independent Set: Self-Training through Dynamic Programming

Advances in Neural Information Processing Systems (NeurIPS)

This work presents a novel graph neural network (GNN) framework for solving the maximum independent set (MIS) inspired by dynamic programming (DP).

Multi-Agent Reinforcement Learning with Graph Neural Networks for Online Multi-Hoist Scheduling

Bachelor's Thesis

This thesis explores an approach to solving the online multi-hoist scheduling problem by combining graph neural networks and multi-agent reinforcement learning.

Public Relations Manager

EPFL Google Developer Student Club

Bridging the gap between theory and practice by giving the students a platform to develop their projects and participate in workshops, seminars, hackathons, and more

Student Representative

Department of Advanced Computing Sciences

One of two student representatives of 800, responsible for raising students' issues, ensuring program quality by renewing study material, extending the curriculum by adding new courses, and looking into internal regulations of the bachelor program

Board Member

MSV Incognito

Treasurer, secretary, and community manager at a department-wide study association comprised of 800 members

Tutor (Co-Founder)

Inforca

Tutoring and hosting group exam-preparation sessions for undergraduate students in computer science- and mathematics-related topics