AMIR PARIZI

• LLM English-SQL Bot: Enhanced the performance of the LLM (Azure OpenAI - GPT 3.5) model by optimizing it through Snowflake database tuning. This optimization facilitated natural language database searching in Snowpark for the BI team.

Tools : Azure OpenAI , Python , Snowpark , Streamlit , Snowflake, GPT 3.5

• Customer Segmentation 360 Dashboard: Developed a Customer Segmentation algorithm for a massive 900M transaction dataset, resulting in improved targeted marketing strategies and the creation of the "Customer 360 Dashboard" for Staples US stores.

Tools : Python , SQL , Snowpark , Clustering , Big Data

• A/B Testing: Streamlined marketing A/B testing processes with a Python application, contributing to enhanced testing procedures.

Tools : Python , A/B Test , Bayesian Probability , Flask

• Sentiment Analysis: Conducted Sentiment Analysis on a dataset consisting of 560K survey responses, achieving an impressive accuracy rate of 96.7% in labeling by employing TF-IDF and LSTM models.

Tools : Python , Snowflake , TF-IDF , NLTK , PySpark , Tableau , PowerBI

• Time Series Forecasting: Provided mentorship to a Junior Data Scientist in the realm of Time Series Analysis and successfully implemented a production-ready LSTM model in Snowflake. This model enabled the generation of hourly flash reports and the detection of data anomalies.

Tools : Python , Snowflake , Neural Network (Keras) , Pytorch , Time Series Analysis ( SARIMAX ) , Tableau

• Life Time Value Model: Transformed the Life Time Value (LTV) model from Pandas to Pyspark, dramatically reducing the training time from 2 days to just 40 minutes for modeling 17M customers using the XGBoost algorithm.

Tools : Python , Snowflake , XGBoost, Pyspark

• Anomaly Detection in Staples IIS Logs: Leveraged Azure's built-in anomaly detection feature and Deep Learning models (RNN and LSTM) to identify anomalies in Staples IIS logs, thereby preventing prolonged error experiences on the Staples website

Tools: Python , Neural Network (LSTM) , Tableau , Azure DevOps , Django

• Quality Check Function for Staples.com Upload Page: Developed a quality check function for the Staples.com upload page to ensure business cards met border criteria and prevented information cutoff during printing, achieving a precision rate of 91%.

Tools: Python , Neural Network (LSTM) , Tableau , Azure DevOps

• Data Migration: Successfully migrated data from MySQL and DataBricks to the Snowflake and Snowpark environment.

Tools: Python , Snowflake , MySQL , DataBricks, SnowPark

• NLP/Computer Vision Analysis in Fashion Industry: Conducted NLP and Computer Vision analyses on various fashion industry products to create machine learning models for consulting purposes, assisting new retailers.

Tools: Python , MySQL , Tf-IDF , NLTK , Deep Learning (LSTM)