AI / ML researcher & engineer
Mohammadreza
Ahmaditeshnizi
I build AI products and train the models inside them. I care most about the work that holds up once real people start using it.
About
I'm an AI and machine-learning engineer and researcher. I'm finishing a master's in Computer Science (AI) at Leiden University, after a bachelor's in Computer Engineering at Sharif University in Tehran.
Most of my time goes to two startups. I'm a co-founder of StudyPuff, a study app for students, and I build the product and the models inside it. I founded AutoLabKimia, which runs quality control for chemistry labs. Alongside that, I'm a research assistant at Leiden, and I research deep learning and reinforcement learning, with four published papers.
Honors
- 2019Rank 72 nationwideNational University Entrance Exam (Konkur), mathematics and physics group
- 2018National Scientific EliteIran's National Elites Foundation (INEF)
- 2017Gold Medal & Puzzle WinnerInternational Math Competition (IMC), National Stage
- 2014 to 2016Gold / Silver / Bronze MedalsIran's National Junior Math Olympiad
Selected work
All work →
StudyPuff
Co-Founder, AI Engineer & Full-Stack Developer
Since 2026
A live study platform where students join others who are already studying, built on Next.js and Supabase, on the web and in iOS and Android apps. The team was selected for Venture Academy at PLNT, Leiden's startup incubator. Our founder, Elaine Herkul, runs the StudyWithElaine YouTube channel, which has 17.1k subscribers. I'm now building its in-app AI study assistant, a RAG pipeline that embeds and semantically searches students' notes with sentence-transformers and pgvector, plus a recommender that matches students to study partners and sessions.
AutoLabKimia
Founder
Since 2025
PreciseICP, a full-stack SaaS platform that automates quality control for geochemistry labs running ICP-OES. It handles wavelength selection from a certified reference-material library, calibration-regression flagging, dilution-corrected concentrations, and per-batch validation against OREAS standards, with asynchronous batch processing and subscription billing. In production at 30+ commercial laboratories.