Ahmed – DBT, Databricks, Python

A TV audience-measurement and advertising data platform that connects fragmented viewership, device and identity data so advertisers, agencies and media networks can reach and measure the same households across smart TVs, mobile and web. Its core is a household identity graph built from IP addresses, hashed email addresses, mobile ad IDs, and connected-TV identifiers, which powers audience segmentation, cross-device targeting, and campaign attribution. The platform must also honor consumer privacy requests (such as CCPA deletions) across all linked identifiers and data sources.

• Designed and built the CCPA deletion pipeline end-to-end in Airflow: ingests hashed-email delete requests, maintains an upsert-only weekly hash index so repeat requests are recognized, fans out from each household to all linked identifiers, validates removals and writes a per-row proof record so every deletion is audit-traceable.
• Built household identity-graph datasets from IP, hashed email, mobile ad ID, and CTV signals with weighted link strength, handling edge cases such as shared Wi-Fi, high-traffic IPs, and identifiers linked across households, to feed probabilistic matching models.
• Built Spark/Databricks feature-engineering pipelines for behavioral segmentation and attribution modeling.
• Optimized large-scale distributed joins and Delta tables (OPTIMIZE, liquid clustering, partition pruning) for model-training and reporting workloads.
• Implemented governed data models with lineage tracking that power business-critical reporting for cross-functional teams.

A search-marketing analytics platform for a global home-furnishing retailer operating online stores in dozens of markets. It consolidates organic and paid search data from Google with internal web and sales data to estimate how much revenue search drives, classify search queries by intent, and guide marketing investment. The platform feeds a control tower of dashboards used by marketing and e-commerce stakeholders to track search performance across markets.

• Redesigned the search revenue estimation models using historical and behavioral data and delivered control-tower dashboard models for multiple stakeholder groups, working within a dbt + BigQuery-only setup with no orchestration or CI tooling.
• Simplified and rationalized the search data model and refined query-classification logic, engineering historical behavioral datasets for ML-based query classification.
• Modernized ELT pipelines in dbt on BigQuery for scalability, cost efficiency, and stability, and integrated search data with internal and external sources.
• Implemented automated data-quality checks and monitoring across the search data pipelines.
• Built feature pipelines for marketing attribution and predictive revenue modeling, and integrated ML outputs into BI-ready data layers.
• Promoted data-mesh practices, documented the new models, and trained analysts and stakeholders on using them.

A partner-data platform for one of the world's largest online travel marketplaces, serving the hotels, apartments, and other accommodation providers listed on it. It turns booking, performance, and service data into governed data products that power partner-facing reporting, partner performance scoring, churn prediction, and segmentation. The platform also provides a shared landing zone where incoming data is checked for quality and freshness, and sensitive fields are masked before use.

• Led the migration of Partner Services data workloads from legacy Hive/Hadoop to Snowflake and introduced dbt for modular, tested transformation workflows.
• Replaced several point tools with a single landing-zone platform handling data-quality checks, freshness monitoring, and data masking, simplifying long-term maintenance.
• Acted as project and policy owner for a team of data analysts, setting data-modeling standards for scalable, governed partner-facing data products.
• Built feature-store-style curated layers for partner performance scoring models and pipelines enabling churn prediction and partner segmentation.
• Optimized Snowflake workloads for model-training efficiency and integrated ML outputs into operational reporting datasets.
• Aligned stakeholders on data design and architecture decisions across engineering, analytics, and partner-facing teams.

A suite of AI-driven marketing optimization tools built by a global advertising and media agency network for its brand clients and media planners. The tools feed client marketing data into machine-learning models and serve their predictions and recommendations through web applications, helping marketing teams decide how to optimize campaigns and media spend.

• Built API pipelines that integrated ML model inference into the backend services powering AI-driven marketing optimization tools.
• Implemented data-preprocessing workflows that prepared client marketing data for AI experimentation.
• Optimized Python services handling model input/output transformations between the models and the application layer.
• Developed React front-end features on top of the Python backend, delivering both sides of new product functionality.

Internal compliance and product-security tooling for a world-leading manufacturer of semiconductor lithography equipment. It collects compliance, security, and audit data from engineering and product systems into structured, traceable datasets. These datasets support regulatory and security audits, security analytics, and automated risk and anomaly detection across the company's products.

• Served as Technical Product Owner for the compliance data tooling, owning the backlog, priorities, and delivery while staying hands-on in the code.
• Built Python ETL pipelines on Azure that turned compliance data into structured datasets, enabling AI-based risk-detection automation.
• Built scalable ingestion pipelines supporting anomaly-detection initiatives and engineered audit-trace data layers for advanced analytics.
• Set up and maintained CI/CD workflows across Jenkins, Bamboo, GitLab CI/CD, and GitHub Actions, with Docker-based builds and Bash automation.
• Improved pipeline reliability for security-analytics workloads and acted as Digital Ambassador, driving adoption of new tooling across teams.

A cloud data platform for a media and publishing group running a portfolio of digital news and content properties. It consolidates content, audience, and transactional data from many internal and external sources into a data lake and warehouse. The platform powers business intelligence and reporting, behavioral analytics, content recommendations, and near-real-time personalization for readers.

• Built the group's data lake and warehouse architecture across AWS and GCP, provisioned with Terraform and orchestrated with Airflow.
• Built and ran data pipelines ingesting from multiple internal and external sources into Redshift and RDS, and managed the underlying AWS infrastructure.
• Built ML-ready data marts for behavioral analytics and recommendation systems.
• Engineered streaming pipelines supporting near real-time personalization models.
• Implemented validation checks to reduce ML feature drift and partnered with data scientists on feature-engineering pipelines.