Python Back-end Developer2021 - PRESENTSciMar One, LLC
Technologies: Python, Azure, Version Control, APIs, Databases, MongoDB, SQL, Pyodbc, Async, Web Scraping, Object-oriented Programming (OOP)
- Developed the core logic of the website back end using Python and hosted on Azure.
- Designed the database architecture for the website back end.
- Designed several Azure Logic apps for curation functionality.
Senior Data Scientist2018 - PRESENTFreelance
Technologies: Amazon Web Services (AWS), OpenCV, NLTK, Keras, Computer Vision, Data Analysis, TensorFlow, Data Science, Git, Jupyter, Pandas, AWS, Machine Learning, Writing & Editing, Generative Adversarial Networks (GANs), Business Process Optimization, Natural Language Processing (NLP), Artificial Intelligence (AI), Deep Learning, Tableau, R, Python
- Developed a program that takes a lung CT scan as an input and gives back numerical and visualization output of detected nodules in the scan, along with malignancy score, on the nodule level and patient level.
- Developed a haircut and eyelashes recommendation system based on facial features and eye features; extended it to generate the photo of a human with the recommended changes, using GANs.
- Developed a receipt OCR and text classifier to categorize the items inside the receipt to various classes.
- Implemented a community detection algorithm, for academic purposes, to detect better design patterns for the UK railway stations.
- Designed and built a tremor classification model based on 3D gyroscope acceleration readings.
- Developed a snoring detection program built for medical purposes, this was built on AWS Sagemaker.
- Built an ace percentage forecast model to forecast the number of aces that will be made by a player in ATP or WTP; it reached 0.4 total loss in it.
- Developed a resume parser in Spanish and English, trained and deployed using AWS instances.
- Developed a sentiment scoring model based on restaurant reviews (Yelp dataset).
- Developed the RL agent based on Deep Q-Learning, to play Sichuan mahjong, along with creating a rule-based model to generate data acting as a start point performance, used SL to train on the generated data, and then RL to enhance the performance.
Computer Vision / Deep Learning Engineer2020 - 2021Sports Vision Lab
Technologies: Python, Deep Learning, Machine Learning, Computer Vision, Version Control, Tracking, OpenCV, PyTorch, TensorFlow, Keras
- Developed a soccer field registration module using Synthetic edge map dataset creation.
- Developed a Player detection and tracking module using Deep Sort and Yolo.
- Developed a Jersey number recognition module using super-resolution gans and deep learning.
- Developed an unsupervised model to classify each player's team.
Data Scientist2018 - 2019Synapse Analytics
Technologies: OpenCV, Keras, TensorFlow, Data Science, Git, Jupyter, Pandas, Machine Learning, Data Analysis, Data Visualization, Natural Language Processing (NLP), Computer Vision, Deep Learning, Tableau, R, Python
- Developed a cement market price daily forecasting, for a big cement firm, to reduce the loss of money between company and distributors, with ten days horizon; 91% of forecast values were within a 5% error margin.
- Implemented a store assortment forecast to forecast the weekly demand of products and give the best combination of products to get the highest revenue possible; 90% of forecasts were within a 5% error margin. Built on an AWS EC2 cloud instance.
- Developed a lot of presentations and dashboards for various projects using Superset, Tableau, and plotly.
- Handled big databases, and maintained its structure, design, and data flow.
- Worked on a clinic recommendation system, along with a time series forecasting model to predict when will be the next visit for a patient.