Papers first, then the things people pay for.

Labs

3 labs at Sharif.

Computer vision, then synthetic data, then reinforcement learning.

1. Research AssistantJun 2024 — Feb 2025

   HABIBI Laboratory, Sharif

   Supervisor: Dr. Jafar Habibi

   • Combined ensemble feature extraction with PPO-based RL for software defect prediction.
   • Co-authored arXiv:2412.07927, the SDPERL paper.
   RLPPODefect predictionEnsemble methods
2. Research AssistantSep 2023 — Nov 2024

   RADIAN Laboratory, Sharif

   Supervisor: Dr. Bardia Safaei

   • Built synthetic data pipelines using GANs, VAEs, and Gaussian mixture models for WiFi-based crowd-size estimation.
   • Improved out-of-distribution performance by 200% on unseen environments.
   GANsVAEsSynthetic dataWiFi sensingOOD
3. Research AssistantAug 2023 — May 2024

   RIML Laboratory, Sharif

   Supervisor: Dr. Mohammad Hossein Rohban

   • CNN-based counterfeit black-tea detection from patched and unpatched imagery, reaching 95% accuracy. Co-authored a paper in MDPI Horticulturae (2024).
   • Led a team using 3D CNNs on MRI scans of beans for non-destructive quality assessment.
   • Created 50 hours of video tutorials and 400 pages of lecture notes on supervised, deep, and probabilistic learning.
   • Built a 21-server tunnel VPN with automated failover that sustained uninterrupted lab connectivity through nationwide internet shutdowns.
   CNNs3D CNNsComputer visionTeachingInfrastructure