Mohammad Mazraeh

Working on AI based Code Generation at enterprise scale code bases in MongoDB's App Modernization Platform group.

• Contributing to AutoGen, an Open Source, event driven, agentic AI framework to interact with Gen. AI, LLMs, etc. • Built a ChatBot using Azure AI Services on technical documents and previous incidents to improve the on-call experience in the team • Multiple initiatives on the Microsoft Experimentation platform to help Microsoft Office teams to experiment and iterate faster • Advocated to improve team's engineering standards including monitoring and on-call process and engineering excellence to use Infrastructure as Code (Bicep) for resource provisioning.

• Delivered 1000+ customer facing ML models/week by leading development of CI/CD/CT MLOps pipelines on AWS. • Developed sales estimator models using PyTorch, MLFlow, ECS, andSageMaker. • Enabled keyword search volume forecasting feature by training ML models in Darts library and deploying in AWS SageMaker. • Architected and lead the effort to migrate the workflows from a single account to Cross-Account resource provisioning model on AWS using CDK as Infrastructure as Code. • Reduced #ML models trained on bad data by adding automatic data quality monitoring using GreatExpectations. • Hired, trained, and managed a team of 10 data scientists and ML engineers. • Created tools to help Sales/CSM explain data science model outputs to customers.

• Enabled two-way communication with Autonomous Robots in intermittent connectivity environments by implementing a scalable event-driven serverless architecture as company’s IoT solution. • Helped the team to build proper CI/CD pipelines to build CPU and GPU enabled docker images for various software deployments on the fleet of robots

• Developed a price-based demand forecast model for price optimization using Python and Spark on Azure Databricks

• Implemented Natural Language Processing algorithms for gender recognition and quote extraction using SpaCy and NLTK in Python and MongoDB for the Gender Gap Tracker project.

• Reduced 30% of call center costs by Implementing Cart Auto Confirmation ML Model in across-functional team using the in Java using a semi-supervised approach. • Led a team of 4 in developing a container-based big data platform to track users and deliveries in real-time using Scala, Apache Spark, Kafka, Cassandra, and YARN. • Enhanced product search by optimizing elasticsearch cluster and queries and adding Auto-complete and Did You Mean features. • Designed and developed the first big data platform for the company to analyse realtime clickstream data using Apache Spark, Apache Kafka, Elasticsearch and Java Script.