MAHMUD - UN - NABI

Data Engineer II | Optimizely Bangladesh

January 2024 - Present

• 1.Leading the Annual Recurring Revenue audit automation project from engineering perspective that resulted in almost 75% reduction in manual audit efforts as well as improvised automated audit process.
• 2. Authored DBT dags related to audit automation process in Snowflake, orchestrated in kubernetes.
• 3. Optimized the existing reverse batch ETL pipelines that ingest over 2 million records daily from Snowflake to Salesforce.

Data Engineer | Optimizely Bangladesh

June 2022 - Present

• 1. Containerized and implemented robust, modular data pipelines with GitHub Actions, AWS S3, AWS ECR, AWS EKS.
• 2. Migrated and reduced memory consumption of existing legacy data pipelines by 55% built in Postgres (in form of stored procedure scheduled as cronjobs) data warehouse into new tech stack (python, Snowflake, DBT, Argo Workflow).
• 3. Implemented scalable cloud-friendly and self-healing data flow pattern for reverse ETL purpose.
• 4. Orchestrated batch data pipelines using python, Apache Airflow (MWAA), Argo Workflow (AWS EKS) to ingest data into the main data warehouse (Snowflake).
• 5. Implemented data quality monitoring product for the internal Sales Ops team.

Junior Software Engineer (Data Science Team) | Shohoz Limited

January 2021 – June 2022

• 1. Worked on building the initial phase of food recommendation engine. The task was to approximate similar food items given a name of food items using ANN (approximate nearest neighbor) developed by Spotify. It was used for candidate generation based on the similarity of added items in the food cart.
• 2. Built and maintained ETL pipelines orchestrated by Luigi and containerized by Docker.
• 3. Worked on building ETL pipelines to ingest semi-structured data from MongoDB to Amazon Redshift.

Intern Software Engineer | Shohoz Limited

October 2020 – January 2021

• 1. Built and pre-processed dataset for the food recommendation engine. Helped QA engineers to test out the food recommendation pipeline using long tail plot analysis to segment the most popular and non-popular items.