Sina Motevalli Bashi Email & Phone Number

Canada

Toronto, On, Ca

ML Training/Deployment for programmatic advertising

Toronto, On, Ca

Seattle, Wa, Us

Worked on several data-intensive ETL pipelines computing various advertising performance metrics based on high volume of daily facts (Spark, Scala, EMR, Orchestration, ...)Worked on server-less (Lambda-based) APIs that consume data from NoSql databases with extremely high traffic and low latency. The Databases are often synced with the output of daily ETL jobs. (Lambda, DynamoDB, StepFunctions, Python, Typescript, ...)Worked on DynamoDB schema design and implementation for various projects.Became an Amazon security certifier and certified the Alexa Conversation management service prior to Launch. The certification process entails various types of high-standard auditing and inspection of the implemented system to discover security vulnerabilities and fixing them before launch.Worked on design and implementation of a federated-architecture based system by which various teams can independently design and deploy front-end widgets and pages to Amazon advertising website. Won Amazon advertising spotlight award for this with my team members.Got heavily exposed to AWS and infrastructure as code, in particular wrote a lot of cdk code in typescript and became very comfortable with the relevant concepts.My 2021 Hackathon team was placed 1st in our advertising org.My 2022 Hackathon team was placed 2nd in our advertising org.

Toronto, Ontario, Ca

Tools and Tchnologies used: Docker, AWS, Kafka, Spring boot, Fastapi, Kubernetes, Openshift, Jenkins, Microservice architectureWorked on multiple NLP-based machine learning projects.Research exposure: Various NLP areas: word2vec, named entity recognition, dependency parsing, seq-to-seq language models, BERT, semantic parsing.Got proper exposure to microservice architecture (by working on multiple ML applications that were built as a collection of inter-connected microservices)Implemented various end-points and authentication filters in a spring boot application.Implemented between service message passing in a microservices application for gathering metrics using Kafka and displaying them in AWS cloudwatch.Created a fully dockerized stack on AWS for a microservice application that utilizes GPU for training and inference in the model service.Created fastapi-based web services in python and deployed them in a Kubernetes Cluster (Openshift) with jenkins and urban code deploy.

Toronto, Ontario, Ca

Helped increase the accuracy of our indoor positioning system by experimenting with and implementing various machine learning algorithms including locally preserving PCA, edited KNN, Clustering, EM, Decision trees, ..Research: Did literature review on indoor localization, got proper exposure to metric learning, manifold learning and dimensionality reductionDesigned and implemented algorithms for computing the top-paths traversed in indoor environments (used graph-theoretic and mathematical approaches)Implemented our self-serve user creation functionality for different user authorizations and page accesses.Implemented a System for filtering departing/arriving passengers and producing detailed passenger behaviour analysis.Implemented various API services and Utility classes in Java for computing a variety of metrics related to customer/passenger behaviour based on our indoor positioning system. (Including realtime metrics and predictions)Wrote scripts for detection of sources and causes of data discrepancy which led to illuminating insights.