Anders Kiss

Data Scientist 10/2022 to Present

Gemological Institute of America (GIA) – remote

• Developed deep learning and machine learning models for diamond grading purposes based on image classification and semantic segmentation algorithms
• Applied cloud based platforms (e.g. AWS Sagemaker and S3) for deployment of scalable solutions
• Utilized statistical programming tools to perform advanced analytics and visualizations
• Worked with gemologists and researchers on experiments and statistical analyses for root-cause investigations

Medical Device Consultant 11/2020 to 09/2021 

Varian – remote

• Licensed the nanoparticles in the European Union (CE-Mark registration) collaborating alongside Engineering, Operations, Marketing, Product Management, and Sales departments
• Managed the technical data and product dossiers for embolization nanoparticles (oncology product)
• Led international registrations for the microspheres globally

Medical Device Consultant 09/2019 to 03/2020 

Medtronic – remote

• Submitted European Medical Device Regulation (MDR) technical file (CE-Mark) dossiers for nerve-monitoring spinal product
• Provided regulatory support in New Product Development meetings pertaining to IEC 60601-1 Basic Safety and Essential Performance, IEC 60601-1-2 EMC, IEC 62366 Usability, IEC 62304 Software deliverables

Master of Science, Artificial Intelligence 09/2021 to 09/2022 

Dublin Business School - Dublin, Ireland

// Coursework

• Machine Learning & Pattern Recognition, Deep Learning, Reinforcement Learning, Recommender Systems, Natural Language Processing (NLP), Programming for Data Analysis, Graph Theory, Cognitive & Ethical Dimensions of Artificial Intelligence

// Dissertation

• A comparison study of LSTM and GRU algorithms and a feature analysis of multivariate macroeconomic features pertaining to their capacity to improve accuracy of Bitcoin price 

// Projects

• A data acquisition / preprocessing pipeline script to obtain stock market data, create technical indicator, with data persisting to a database (Microsoft SQL Server). 
• A Content-based recommender system (RS) script is for a content-based recommender system (RS) that I built. This basic system recommends an Airbnb listing to a User based on the description fields of a their historical stays. The code implementation compares recommendation accuracy of the BERT and TF-IDF algorithms 
• An NLP topic modeling application to better understand the major themes for that day’s news headlines by ingesting the Python API, NewsAPI. It can be found here: 
• A Long Short-Term Memory (LSTM) versus Gated Recurrent Unit (GRU) Neural Network predictive, financial model in Keras/Tensorflow that ingests Python’s yfinance API to predict tomorrow’s price.
• A Python script that utilizes that targets communities from the MovieLens database (containing ~38,000 movies) via the graph theory (machine learning) algorithms, Label Propagation (Community Detection Algorithm), and K-Nearest Neighbors (Similarity Algorithm). 
• A Tensorflow / Keras image classification model compared the accuracy of three image classification models: ResNet50, VGG16, and a custom deep learning, neural network. For each model above, first a baseline model was trained and evaluated. Afterwards, a dropout layer was added to each model and compared in terms of the baseline models.