Hadi Davardoust
๐ง davardoust@gmail.com
๐ +989353449388
๐ผ LinkedIn
๐ Google Scholar
๐ Education
University of Tehran | Nov 2020 - Feb 2023
Masterโs degree in Network Science and Technology, GPA: 4/4 (18.52/20)
- ๐ Thesis: Hate speech classification in social networks using machine learning models, supervised by Dr. Hadi Zare
- ๐ Selected Courses: Machine Learning and Data Mining (18.25/20), Simulation and Modeling (20/20), Cellular Automation and Complexity (18.5/20), Advanced Network Science (18.3/20), Mathematical Theory of Systems (18.52/20), Special Topics (19.74/20)
Shahid Madani University | Sep 2007 - March 2012
Bachelorโs degree in Information Technology engineering
- ๐ป Capstone Project: Building a Java SMS platform using Bluetooth for message delivery
๐ฌ Research Interests
| ย |
ย |
ย |
| ๐ง Large Language Models |
๐ฅ Healthcare Applications |
๐ Anomaly Detection |
| ๐ฉป Medical Imaging |
๐ธ๏ธ Graph Neural Networks |
๐ Network Analysis |
| ๐ Medical Data Representation |
๐ Transfer Learning |
๐๏ธ Computer Vision |
๐ ๏ธ Skills
- ๐ป Programming: Python; SQL; JavaScript; C/C++; Bash
- ๐ค LLM Technologies: LangChain; RAG; MCP; OpenAI API; Vector Databases
- ๐ Data Processing: Pandas; NumPy; SciPy; Matplotlib; Jupyter
- ๐ฃ๏ธ Language Proficiency: Duolingo English Test: 130 (Dec 2024)
๐ผ Work Experience
๐ข Saman Insurance | Jun 2024 - Present
Machine Learning Engineer | Tehran, Iran
- ๐ Built a retrieval system processing 10M+ medical documents annually, measuring retrieval reliability (90% confidence) versus insufficient evidence cases
- ๐ Developed document classifier categorizing medical records into 8 types (prescriptions, lab results, doctor reports) with 68% accuracy, flagging uncertain predictions for human review
- โ ๏ธ Designed anomaly detection tools flagging unusual/incomplete patient-submitted documents during medical claim processing
- ๐ Created Iranโs first fully online health insurance platform with real-time policy visibility, revealing gaps in patient records for certain subgroups
- ๐ค Collaborated on automated claim triage system identifying cases requiring human adjudication due to insufficient/conflicting data
๐ Mofid Securities | Jan 2022 - Jun 2024
Machine Learning Engineer | Tehran, Iran
- ๐ฌ Improved sentiment analysis for call transcripts, boosting customer satisfaction detection accuracy from 83.2% to 91.1%
- ๐ Developed LDA-based text summarization system for call transcripts, expediting customer pain point identification
- ๐ญ Implemented Liveness Detection algorithms for e-KYC systems using facial recognition, motion analysis, and biometric data
- ๐ค Enhanced Face Verification accuracy by optimizing FaceNet algorithm (89.32% โ 95.34%)
- ๐ณ Utilized Docker for e-KYC projects, ensuring consistent environments and streamlined deployment
๐ป Fanava | June 2012 - Dec 2021
Software Engineer | Tehran, Iran
- ๐ Developed predictive model for personnel exit time estimation, enabling proactive resource planning
- ๐ฎ Improved demand forecasting accuracy using advanced ML algorithms
- ๐ฅ Designed comprehensive HR solutions for 800+ personnel including insurance and benefits management
- ๐๏ธ Developed data warehouse solution with ETL processes, integrating data from existing ERP systems
- ๐ Implemented BI reports using Power BI for interactive visualizations and data-driven decisions
- ๐จโ๐ผ Created CEO-level HR analytics dashboards (turnover, payrolls, healthcare, work hours)
- ๐ Developed Project Management system integrated with Microsoft Project Server
- โ
Developed Quality Assurance Management system
- โฑ๏ธ Implemented tailored timesheet and tasksheet solutions across departments
- ๐ฏ Led recruitment efforts, designing interview questions and code challenge assessments
- ๐จโ๐ซ Hired and mentored a team of 5 developers from top Iranian universities
๐ฌ Research Experience
๐ฅ University of Tehran, DSP Lab | Sep 2022 - Present
Research Assistant | Tehran, Iran
- ๐ฉบ Contributed to Persian Medical Question Answering System development, applying advanced NLP techniques for accurate information retrieval and response generation
๐งฌ University of Tehran, Kavosh Lab | Sep 2021 - Present
Research Assistant | Tehran, Iran
- โ๏ธ Researched and implemented novel Graph Neural Network architectures (including Graph Transformers) for molecular property prediction, improving MAE by 4.5% over previous best model
๐จโ๐ซ Teaching Experience
-
๐ PyTorch Tutorial | University of Tehran | April 2023 - July 2023
Instructed PyTorch workshops for students working on dissertation projects
-
๐ Introduction to Python | University of Tehran | Sep 2022 - Dec 2022
Instructed introductory Python courses for new students in faculty forums
-
๐ Simulation and Modeling (TA) | University of Tehran | March 2022 - July 2022
Facilitated course through homework correction and unwind sessions
-
๐ค Machine Learning and Pattern Recognition (TA) | University of Tehran | Sep 2021 - March 2022
Created and taught neural network lessons including plans, unwind sessions, and homework correction
๐ Papers
-
๐ H. Davardoust, H. Zare, H. Rafiee Zade, โThe Dark Side of Instagram: A Large Dataset for Identifying Persian Harmful Commentsโ โ submitted to SoCal NLP Symposium 2024
-
๐ H. Rafiee Zade, H. Zare, M. Ghassemi Parsa, H. Davardoust, M. Shariat Bagheri, โDCOR: Anomaly Detection in Attributed Networks via Dual Contrastive Learning Reconstructionโ โ Thirteenth International Conference on Complex Networks and Their Applications 2024
-
๐ H. Davardoust, H. Zare, โComplex network analysis of keywords co-occurrence in Graph Neural Networkโ โ Learning on Graphs Conference 2025
๐ Honors
- ๐ฅ Ranked in top 0.5% in National University Entrance Examination for Masterโs Programs in Computer Engineering (2020)
- ๐ฅ 2nd Place in Masterโs Program in Network Science and Technology at University of Tehran (2023), based on cumulative GPA
- ๐
Received Outstanding Masterโs Thesis Award for best thesis in the program (2023), achieving highest possible score in defense