Ron Medina

Æ +63-995-***-**** Q adh1ec@r.postjobfree.com

linkedin.com/in/ron-medina-5a411218a/ particle1331 With a strong background in mathematics, my goal is to develop machine and deep learning systems that would (1) optimize business functions and (2) lead to valuable information and business insight. Education

University of the Philippines - Diliman June 2013 – Dec 2018 Bachelor of Science

Major in Mathematics

Relevant Courses:

+ CS11 Intro. to Computer Programming 1 (Python) - 1.00

+ Math171 Numerical Analysis - 1.75

+ Math123.2 Advanced Calculus 2 - 1.50

+ Math110.2 Abstract Algebra 2 (Linear Algebra) - 1.25 Eskwelabs July 2019 – Oct 2019

Data Science Bootcamp

Attended a 10-week bootcamp which included 160 hours of in-class learning in addition to coursework. At the end of the bootcamp, I resented my capstone project in front of industry leaders in Makati City, Philippines.

Skills

Data analysis

+ Data visualization, data wrangling, and exploratory data analysis of datasets (<5M) using Pandas, seaborn, matplotlib, and NumPy.

+ SQL, probability modelling, basic statistics, scikit-learn Machine Learning

+ Training and tuning of deep neural networks in Keras, PyTorch

+ Training and tuning of machine learning models in scikit-learn (SVM, RF, logistic regression, k-Means, Naive Bayes, PCA, etc.)

+ Recommender systems (CF, next-item sequential), fraud/anomaly detection (highly imbalanced data)

+ Gradient boosting (Catboost, XGBoost, LightGBM), ensemble learning (stacking) Model deployment

+ Experience with deploying models in Django

+ Version control using git

+ Familiarity with bash scripts

Projects

Data-driven reconstruction of a chaotic dynamical system Oct 2019

+ Code is written in Tensorflow 1.x.

+ Sampled data from the Lorenz (1963) equations is fed into a dense MLP network.

+ Obtained good results for correlation dimension and Lyapunov exponent in the reconstructed system. Recommendation system May 2020-Present

+ Recommendation system for online food delivery (simlar to foodpanda). Currently deployed on four different franchises, with multiple stores per franchise. Trained over 5M transactions.

+ Collaborative filtering, item-item based recommender model

+ Code written using numpy, pandas, and scipy (sparse matrices)

+ Increased ATC=3.38 (ave. ticket count), ATV=$11.49 (ave. ticket value) to ATC=4.86, ATV=$14.21 in the first two months in production.

+ An improved version of the recommender currently under development is based on [Tang & Wang (2018)] uses a convolutional network over embeddings of sequences.

+ Predicts within a Django server.

Fraud detection July 2020-Present

+ Uses a stacked CatBoost, Imbalanced-XGBoost basemodels, and Logistic Regression model to detect fraudulent transactions.

+ Prediction time is 1.00 sec, with a true positive rate of 88% and a true negative rate of 98% on a test set of 100,000 transactions which drastically improved previous rule-based approaches.

+ Prediction is done using a Django server.

+ Currently being integrated into existing systems. Employment History

Machine Learning Engineer Mar 2020 – Present

+ Currently a Machine Learning Engineer at BrewedLogic, Inc.

+ Preprocessing and analysis of datasets

+ Model selection, training, evaluation (cross-validation) of models

+ Research and implementation of novel techniques and architectures that will be useful to current projects.

+ Deployment of models. Making sure that the machine learning models satisfy the constraints for proper implementation (e.g. size and time constraints).

+ Maintenance of models. Retraining of models as well as monitoring of model performance. Bootcamp Associate Oct 2019 – Jan 2020

+ Developed curriculum materials for the data science bootcamp at Eskwelabs

+ Participated in live Hackatons.

+ Presented a talk at the National Youth Congress held at UP Diliman School of Economics, Nov 2019, on the topic of artificial intelligence and deep learning. Data analyst June 2017 – June 2019

+ Worked part-time at an online shop with 122K followers.

+ Performed basic statistical analysis of shop data

+ Clustered customers based on RFM criteria.

+ Analysis of frequently bought together items using graphs and correlation.

+ Advised proprietor about which items have higher ROI. Tutor Sept 2013 – Dec 2018

+ Tutoring high school students 1-on-1 in mathematics and science.

+ Exam review: NMAT, Civil Service, UPCAT/ACET, etc. Tutor June 2016 – Aug 2016

ACE Tutorial Center, Katipunan, Quezon City

+ Tutored students from Ateneo Grade and High School.

+ Designed curriculum materials, provided homework and project assistance.

Contact this candidate