Henrik Abgaryan

Bio

PhD Student at Lamsade LAB, PSL Dauphine, Paris, France.

I specialize in advanced machine learning techniques for NP-hard combinatorial optimization problems. My research focuses on large language models, reinforcement learning, and generative AI approaches to optimization. My background spans text-to-image synthesis, extensive R&D endeavors, teaching responsibilities, and industry roles in deep learning.

Research Areas: Machine Learning, LLMs, RL for Combinatorial Optimization

Supervisors:

• Tristan Cazenave (tristan.cazenave@lamsade.dauphine.fr)
• Ararat Harutyunyan (ararat.harutyunyan@lamsade.dauphine.fr)

2023 - Present

Publications

• ACCORD: Autoregressive Constraint-satisfying Generation for Combinatorial Optimization
  H. Abgaryan
  Under Review, 2025.
• STARJOB: Dataset for LLM-driven Job Shop Scheduling
  H. Abgaryan, A. Harutyunyan, T. Cazenave
  AAAI25 LLM4Plan (Oral), 2024. [PDF]
• ScheduleExpert: Graph Attention Meets Mixture-of-Experts for JSSP
  H. Abgaryan
  LION19 (Oral), 2023. [PDF]
• LLMs can Schedule
  H. Abgaryan, A. Harutyunyan, T. Cazenave
  arXiv preprint arXiv:2408.06993, 2024. [PDF]
• Randomized Greedy Sampling for JSSP
  H. Abgaryan, T. Cazenave
  The 18th Learning and Intelligent Optimization Conference (LION18), 2025. [PDF]
• House Allocation with Randomly Generated Preference Lists
  H. Abgaryan, A. Harutyunyan, T.N. Le
  2022. [PDF]

Talks

• Presented Randomized Greedy Search for JSSP at the 18th Learning and Intelligent Optimization Conference
  June 11, 2024 | Ischia, Italy
• Speaker at Lamsade Doctoral Spring School
  École du Luat, Écouen-Ezanville, France

Work Experience

• Current PhD student in Paris Dauphine PSL
  Working on intersection of Combinatorial Optimization and AI
  2023 - present
• PICSART
  Machine Learning Scientist at Artificial Intelligence Department
  2021
  • 35+ AI projects in production and R&D.
  • Team lead for text-to-image personalization (few-shot learning); features used by 150 million monthly active users.
  • Working on Text-to-Image Synthesis, Image to Image Synthesis, SOTA model finetuning.
  • Large-Scale Training, GANs and VAEs.
  • Experienced with open-source AI models, particularly in image generation (e.g., Stable Diffusion).
  • Utilized techniques like LoRA for model optimization.
  • PyTorch, TensorFlow production systems.
• Synergy International Systems
  Software Engineer Intern
  2019
  • Backend development using JAVA and Python frameworks.
  • Database development using MS-SQL.
  • Object-Oriented Programming (OOP).

Education

• PhD, Artificial Intelligence
  PSL Dauphine, Paris
  2023 - Present
  • ML for combinatorial optimization
  • Supervisors: Tristan Cazenave, Ararat Harutyunyan
• Master's of Computer Science
  M2 AI Systems and Data, PSL Dauphine
  Paris, France | 2022
  • One of Top 50 of world universities according to Shanghai Ranking.
  • Master's Thesis: "Optimization and Fine Tuning of Large Text to Image Models".
  • PSL Excellence Scholarship recipient
• Bachelor of Computer Science
  American University of Armenia
  2017
  • Literature Survey: Fireflies curing Fault-Tolerance in Robotic Swarm Systems.
  • Literature Survey: Community Detection in Mixed and Temporal Networks.
  • Capstone: Graph Diameter Shrinking Approaches for Pseudo-Random Graphs.
  • National Academic Scholarship recipient

Technical Expertise

• Areas: LLMs, RL, Generative AI, Combinatorial Optimization
• Deep Learning: Model design, large-scale training, parameter-efficient tuning
• Tools: PyTorch, TensorFlow, JAX, Hugging Face, distributed training

Teaching

• Teaching Associate: Introduction to ML
  M1 MIAGE Apprentissage, Paris Dauphine PSL
  September 2024 - February 2025 | Paris, France
• Teaching Associate: Artificial Intelligence
  M1 MIAGE Apprentissage, Paris Dauphine PSL
  January 2024 - May 2025 | Paris, France
• Teaching Associate: Outils en informatique (Excel)
  Paris Dauphine PSL
  January 2024 - May 2025 | Paris, France
• Teaching Associate: Outils en informatique (Internet, HTML)
  Paris Dauphine PSL
  January 2024 - May 2025 | Paris, France
• Teaching Associate: Causal Inference
  American University of Armenia
  January 2022 - May 2022 | Yerevan, Armenia
• Teaching Assistant: Data Visualization
  American University of Armenia
  September 2021 - December 2021 | Yerevan, Armenia

AI News

Latest AI news from around the web