Behrang Koushavand
Verified Expert in Engineering
Machine Learning Developer
Behrang is a geo-statistician, data scientist, and professional engineer with 10+ years of industry experience. He has deep knowledge of multivariate statistics and industry scale optimization algorithms, classical machine learning, deep learning, and computer vision. Behrang has participated in 30+ Kaggle competitions and has experience with different types of data and ML algorithms.
Portfolio
Experience
Availability
Preferred Environment
Python 3, Fortran, Azure, Amazon Web Services (AWS), Python, Keras
The most amazing...
...achievement I've accomplished is to be granted a patent on the application of computer vision in oil and gas reservoir characterization.
Work Experience
AI Consultant
ShotBot Inc.
- Trained multiple models and fine-tuned the workflow to generate professional headshots from user images.
- Reviewed articles, current advancements, and state-of-the-art technologies. Proposed a new workflow yielding superior results compared to existing commercially available solutions.
- Established the workflow and proposed the required infrastructure for the operationalization and commercialization process.
Senior Principal Data Scientist
NielsenIQ
- Contributed to various advanced analytics projects related to geospatial data analytics projects, route-to-market, and precision.
- Implemented a novel method for sales distribution using the linear optimization technique.
- Implemented parallel programming for various existing codes to increase the efficiency of the current methods.
Machine Learning Engineer and Data Scientist
Cenovus Energy
- Developed several ML algorithms and shiny apps deployed to RS-connect, including a deep learning model to classify valid DTS shut-in, linear failure analysis, and predictions of the depth of geological formations using CNN.
- Supervised a data science co-op student on various projects.
- Co-supervised a master's student at the University of Alberta on using ML methods to predict spatial continuity parameters from data.
Python and AI Developer
Syngenta - Cloud Compute
- Integrated computer vision into the internal chatbot, enabling users to import images and ask related questions. Deployed multiple endpoint APIs to Sagemaker for enhanced functionality.
- Built a web scraper for the knowledge base, answered queries using relevant articles, and raised tickets with Selenium, enhancing user support and information access.
- Incorporated a data chat feature into the company's internal chatbot. Users can import multiple data files (CSV, Excel) and pose complex questions for advanced analysis.
AI Developer
Zebra Analytics, LLC
- Trained for TensorFlow to detect different elements from the plan.
- Deployed a model to the GCP server for training and inference.
- Created a web app using Flask and deployed it with a Docker container into a GCP VM.
- Helped and mentored a client with their first cloud solution (GCP).
- Wrote 8 Jupyter notebooks to pre-process the images, train the models, and evaluate the results and a stand-alone notebook to make inferences on new data.
- Helped a client to collect the data and annotate it to achieve their goal.
AI Developer
Zebra Analytics, LLC
- Reviewed literature on the subject of generative plans and layout generation and devised out-of-shelf temporary solutions.
- Ran two inferences using a publicly available trained model on custom data.
- Provided a plan for the next step, guided the customer on how to label their data, and helped them prepare their custom dataset for the next step.
AI Specialist
Oluwadolapo Obatoyinbo
- Defined a few projects for the client for an AI and computational geoscience software prototype and wrote and explained each step. Assisted the client in collecting the data and preparing for the next step.
- Prepared documents and communicated several technical details.
- Presented the team with a clear scope of the work and guided them through different technical processes.
Geo-statistician and Machine Learning Engineer
Husky Energy
- Developed several predictive models for subsurface data using classical machine learning algorithms and a deep learning framework.
- Filed two patents—one has been granted, and the second is in process.
- Led a team of four technical individuals to develop a framework and a dashboard for resource estimation and prediction.
- Built 3D numerical models to characterize reservoir and uncertainty analysis.
- Taught several machine learning courses in the company.
- Received the best presentation award at the Husky tech forum. Presented three new workflows and ML models on subsurface data.
Research Assistant
University of Alberta
- Developed a novel workflow to measure geological uncertainty at long-term production optimization and published several papers on this subject.
- Published several peer-reviewed and conference papers about geostatistical methods in mining and oil and gas engineering.
- Acted as a teacher assistant in a few university courses, including reserve estimation methods.
- Acted as a software developer for Newmont Mining Corp, Denver, USA.
Experience
Rock Permeability Prediction Using Computer Vision
https://patents.justia.com/patent/11182890Later, I built a standalone app deployed at the company for all of the users.
This workflow has been patented.
Computer Vision to Detect Reservoir Changes
https://patents.justia.com/patent/20210080607This is a pending patent.
A Linear Programming Model for Production Optimization with Input Uncertainty
https://www.sciencedirect.com/science/article/pii/S2095268614000792Kaggle Competitions
https://www.kaggle.com/behrangSome of the competitions were:
• Homesite Quote Conversion: Which Customers Will Purchase a Quoted Insurance Plan?
• Rossmann Store Sales: Forecast Sales Using the Store, Promotion, and Competitor Data
• Google Landmark Retrieval $ Recognition 2020
-HuBMAP - Hacking the Kidney: Identify Glomeruli in Human Kidney Tissue Images
-The Winton Stock Market Challenge
-G2Net Gravitational Wave Detection: Find Gravitational Wave Signals from Binary Black Hole Collisions
Education
PhD in Geo-Statistics
University of Alberta - Edmonton, AB, Canada
Certifications
Professional Engineer (PENG)
The Association of Professional Engineers and Geoscientists of Alberta (APEGA)
Skills
Libraries/APIs
TensorFlow, Scikit-learn, Keras, XGBoost, CatBoost, Pandas, Matplotlib, PySpark, PyTorch, PyMongo, OpenCV
Tools
MATLAB, Git, CPLEX, AutoML, GitHub, AutoCAD, You Only Look Once (YOLO), GIS, ChatGPT, Amazon SageMaker, Azure ML Studio
Frameworks
LightGBM, Apache Spark, Spark, Selenium
Paradigms
Linear Programming, Data Science, Parallel Programming, Object-oriented Programming (OOP), ETL
Languages
Python 3, Fortran, Delphi, Python, Visual Basic for Applications (VBA), R, SQL, C
Storage
MySQL, NoSQL, MongoDB, Google Cloud
Platforms
Azure, Databricks, Amazon Web Services (AWS), Google Cloud Platform (GCP), NVIDIA CUDA, H2O Deep Learning Platform, Docker
Other
Computer Vision, Geology, Tkinter, Machine Learning, Deep Learning, Integer Programming, Optimization, Geospatial Analytics, Engineering, Facial Recognition, Convolutional Neural Networks (CNN), Object Detection, Artificial Intelligence (AI), Linear Regression, Linear Optimization, University Teaching, Algorithms, Neural Networks, 3D Image Processing, Datasets, Generative Pre-trained Transformers (GPT), Large Language Models (LLMs), LangChain, GeoPandas, Natural Language Processing (NLP), Cloud Infrastructure, Team Leadership, Cloud, Machine Vision, Image Processing, Algebra, Convex Optimization, Optimization Algorithms, Data Wrangling, Simulations, Physics Simulations, User Interface (UI), 3D, GPT, OR-Tools, Consulting, 3D Modeling, 3D CAD, Graphics Processing Unit (GPU), Vertex, Image Recognition, Writing & Editing, OpenAI GPT-3 API, OpenAI GPT-4 API, Generative Pre-trained Transformer 3 (GPT-3), GPU Computing, Research, Programming, Geospatial Data, Data Engineering, Chatbots, OCR, OpenAI, Minimum Viable Product (MVP), APIs, Generative Adversarial Networks (GANs), Neat, Generative Design, Augmented Reality (AR), Quantum Computing, Frameworks, QGIS, Digital Elevation Models, Satellite Images, Point Clouds
How to Work with Toptal
Toptal matches you directly with global industry experts from our network in hours—not weeks or months.
Share your needs
Choose your talent
Start your risk-free talent trial
Top talent is in high demand.
Start hiring