About Me
I am currently an MSc Data Science student at EPFL and a Research Intern at SwissAI (ETHZ \& EPFL AI Centers). Prior to this, I worked as a Research Assistant at the NLP lab supervised by Prof. Antoine Bosselut.
Education
- MSc in Data Science. EPFL. Lausanne, Switzerland. Sep. 2023 - Present
- BSc in Computer Engineering. Politecnico di Torino. Turin, Italy. Sep. 2019 - Jul. 2023
Relevant Experience
- ML Research Intern. Logitech. Lausanne, Switzerland. Sep. 2025 - Feb. 2026
Computer use; LLM agents. - Summer Research Intern. SwissAI. Zurich, Switzerland. Jun. 2025 - Sep. 2025
Reasoning for vision language models through reinforcement learning. - Student Research Assistant. NLP lab. EPFL, Switzerland. Jun. 2024 - Jun. 2025
Multilingual Pretrain Data Collection[1]; Multimodal Reasoning[a]. - Student Research Assistant. DHLAB. EPFL, Switzerland. Feb. 2024 - Sep. 2024.
Text-to-SQL system; LLM agents development[2]. - Data Analyst. Fater. Italy. Nov. 2022 - May. 2023.
Publications
- ConLID: Supervised Contrastive Learning for Low-Resource Language Identification
Preprint
[paper, code] - LLM-Powered Agents for Navigating Venice’s Historical Cadastre
Preprint
[paper, code]
Teaching Experience
- Student Teaching Assistant. Applied Data Analysis [CS-433]. EPFL. Fall 2024.
Selected Lab Projects
- Multimodal Reasoning. Trained Multimodal Models with Group Relative Policy Optimization for various reasoning tasks. [code, report]
#VLM#PostProcessing#Reasoning#RL - Visual Reasoning with RL. Investigated the extent to which GRPO can be applied to VLMs for visual question answering tasks. [code, report]
#VLM#PostProcessing#Reasoning#RL#GRPO - LLM post-processing (SFT, DPO, RAG). Instruction-Tuned Galactica-1.3B model on Scientific MCQA task using SFT and DPO, and further tuned in the RAG settings. [report]
#DeepLearning #MachineLearning #LLM #FineTuning #RAG #NLP - Large-scale YouTube Analysis. Causal Analysis of Tech channels’ progress on YouTube using the videos published between May 2005 and October 2019. [code, datastory, blog]
#DataAnalysis #CausalAnalysis #MachineLearning #DataVisualization - LLM Fine-Tuning. Fine-tuned 3 LLMs (Mistral-7B, Llama-2-7B, Phi-1.5) on a dataset from X for the stance detection task. [report, blog]
#DeepLearning #MachineLearning #LLM #FineTuning #NLP
External Activities
- Jon and John, YouTube. A YouTube channel I ran with a friend when I was in Italy about Italian Culture, and studying in Italy. Now I sometimes do interviews with the people in AI.
- Jon and John, Telegram. An Uzbek community I ran back in Italy to bring together Uzbek Students in Italy.
- Student Help. An education consultancy service I started with friend. Now run by Muslimjon Nabijonov.
- Jakhongir Saydaliev, Telegram. A personal Telegram channel where I share my thoughts, experiences and knowledge (in Uzbek).
News
| Jun. 2025 | Joined SwissAI to work on reasoning for vision language models through reinforcement learning |
| May. 2025 | Won 2nd place in a hackathon on efficient LLM training [code] |
| Feb. 2025 | Joined a new project on Multimodal Reasoning at the NLP lab |
Show more
| Nov. 2024 | Won the AXA challenge in Lauzhack 2024 Hackathon [project] |
| Sep. 2024 | Participated in DeepFake Hackaton |
| Jun. 2024 | Started Summer Research Internship in NLP Lab on Multilingual Model Training project by Swiss AI Initiative |
| May. 2024 | Presented my LLM Agents project in DHLAB |
| Apr. 2024 | Participated in LLM Hackathon |
| Apr. 2024 | Presented my text-to-SQL project in DHLAB |
| Mar. 2024 | Participated in AMLD 2024 |
| Feb. 2024 | Started Student Research Assistantship in DHLAB on LLM QA system project by Venice Time Machine |