VAHID ASGHARI
- Sai Kung, Hong Kong
Quantitative research professional with a Ph.D. in engineering focused on financial decision-making under uncertainty. I’m a conscientious data scientist with infinite energy, insatiable enthusiasm, and immense perseverance. Over my career, I have demonstrated leadership skills and a history of working in the academic and financial domains.
Education
Sharif University of Technology, Iran
MS in Project management
Sep 2016 - Jul 2018
Sharif University of Technology, Iran
BS in Engineering
Sep 2011 - Jul 2016
Experiences
Hoenir Int. Ltd. - Technical consultancy for startups
Founding Engineer
2021-Present
Partner with Forerunner-data:
- Leading pilot project with two profs. from CityU and ShenzhenU to deploy NLP-based models to novel business research topics.
- Scrapping the Edgar quarterly reports for different companies with Selenium and BeautifulSoup
- Cleaning the raw text and generating the corpus
- Using BERT, fasttext, and other HuggingFace transformers for encoding the raw data
- Training DNN models using PyTorch for predicting corresponding stock price movements
- Evaluating the statistical relationship between reports sentiments and price movements
Partner with Air-Prophet:
- Implemented ETL pipelines and stored the data in a PostgreSQL database
- Trained XGBoost/CatBoost models for air flight duration with 5 minutes MAE for a 6 hours flight
- Reduced the annual costs up to $1.1M and carbon emissions up to 390T for average airline routes
Lifeometer app:
- Led the development using Agile methodology as a Scrum master
- Implemented a scientific health-score module based on statistical concepts
- Designed and implemented the user database on a remote MySQL data server
- Developed REST APIs for the front-end clients using Django/Python
- Deployed the back-end on AWS-EC2 (Linux server), used Nginx as the webserver
- Adapted Test-Driven development methodology. Unit tests covering all use cases and end-points.
RCA-LCA web app:
- Designed and implemented an SQL database for RCA products and machinery
- Created the financial/environmental analysis computation engine with Monte Carlo Simulation
- Developed REST APIs and the backend using Django/Python
- Implemented the front-end client with JavaScript, JQuery, REACT, HTML, CSS
- Deployed the back-end on a Linux server, using Nginx and uwsgi
The Hong Kong Polytechnic University
Post-graduate researcher
2018-2022
ReinforceAM (Thesis):
- Reduced expected costs by 14%, increased utilities by 33% for the Indiana, US, network
- Designed and implemented the first automated multi-agent network-level RL investment agents
- Reduced the computation time of optimization methods from hours to seconds
- Translated the stochastic costs and utility models into python re-usable modules
- Trained ensemble/boosting models with Sci-kit for predicting life cycle costs with a 3% error
- Trained DNN models with Tensorflow/Keras for learning optimization behavior with an 8% error
- Published the first open-source investment valuation model for infrastructure assets
- Enforced clean code for best code readability and maintenance
- Led three post-graduate students in conducting and completing AI-based research
- Published more than 8 scientific papers in top-tier peer-reviewed journals
Hyustan - Forex trading & investment firm
AlgoTrading developer
2016-2018
- Designed a portfolio optimization algorithm (Time-series prediction, Volatility estimation, Reliability analysis, and non-linear optimization) in Python
- Created C++ trading bots in Metatrader based on the requests of the Python backend
- Back-tested >2 million technical indicator-based strategies with Python and C++ in Metatrader
- Successfully returned 61% net profit with an 83% success rate on an allocated fund with biweekly investments in Iran’s stock market within 6 months
Projects
Developed a python library for preprocessing imbalanced data in regression modeling with more than 500 PyPi installations in 6 months, through implementing scientific and mathematical methodologies
Skills: Python, NumPy, PyPi, Pandas
Developed a python library for feature importance analysis on supervised and unsupervised machine learning models through data permutation
Skills: Python, clean code, git, supervised learning, unsupervised learning, K-NN, Naïve Bayes, DNN, Random Forest
CBSA-Wisers Analytics Challenge
Ranked 1st runner-up among more than 20 teams, by 1) developing an NLP-based ML model for prediction of Chinese social media posts’ user engagement with the COVID-19, and 2) finding the significant impact of social media sentiments on the Hang Seng index
Skills: NLP, BERT, fasttext, sentiment analysis, p-value analysis, pandas, NumPy, parallelization
ArchSD/HKSTP Hackathons
Ranked 1st runner-up among 17 teams in Two schemes of this hackathon through 1) training and evaluating ML models for predicting construction costs, and 2) achieving the most accurate time-series prediction for construction costs index
Skills: Deep neural networks, ARIMA, LSTM, statistical analysis
Skills
Data Science Tools
Programming languages: Python (Scikit-learn, Keras, TensorFlow, PyTorch, NumPy, Pandas, SciPy, Matplotlib), C++, Java (Basic)
Database tools: SQL (MySQL, PostgreSQL), MongoDB
Big data tools: Spark, Koala
Machine Learning and Deep Learning
Reinforcement Learning: A2C/DQN/DDPG reinforcement learning
Learning method: Neural Networks, Boosting and other ensemble methods, SVM, Linear regression, Generalized linear models
Deep learning: Natural language processing, Computer vision
Mathematical and Statistical Modelling
Time series modelling: LSTM, ARIMA, VAR
Optimization: Linear programming, Gradient descent methods, Heuristic optimization methods, Convex optimization
Reliability analysis/Simulation: First-order/Second-order reliability analysis, Monte Carlo/Discrete event/Agent-based simulation
Development
Web: Django, JavaScript, CSS, HTML, JQuery, REACT
General: Git, Clean code concepts, Test Driven Development, Scrum project management
Professional and Personal
Personal: Conscientious, Self-motivated, Self-disciplined, Problem solver, Details-oriented
Teamwork: Leadership skills, Effective communicator, Project management skills (with both Waterfall and Agile mindsets), Team player
Publications and Academic Endeavors
All publications are available at: Google Scholar Page
Served as a reviewer of AI/ML and mathematical modelling related scientific papers in:
- Journal of Management in Engineering
- Sustainable Production and Consumption