Shahid Thaika

· Built 70+ of the 110 Airflow DAGs maintained on GCP Composer for a digital advertising company.

· Built Apache Beam / Dataflow pipelines for terabyte-scale workloads and streaming Pub/Sub data.

· Evaluated and prototyped multiple BI platforms before driving adoption of Holisticsas a cost-effective semantic-layer alternative to Looker for scalable enterprise reporting.

· Designed and developed a new star-schema data model in BigQuery, replacing a legacy architecture of hundreds of report-specific tables with about 40 tables and simplifying maintenance.

· Redesigned and optimised legacy ETL queries to reduce processing overhead, pipeline failures, and backfill requirements, and implemented built-in error checks and alerts to improve pipeline reliability.

· Migrated 30+ legacy Looker Studio reports to Holistics, consolidating standalone reports into a centralised, drill-through reporting experience that improved navigation and usability.

· Designed analytical solutions on my own initiative that enabled richer user analysis, expanded segment targeting and predictive modelling population by 6x, and supported adaptation to new PII regulations.

· Built Power BI reports that were embedded in a broader UI to compare predicted vs actual pricing strategy for a retailer with 9,000+ stores. Data sourced from Azure Data Lake and PostgreSQL.

· Benchmarked and optimised a DirectQuery Power BI model and the SQL views powering it, improving load times through reduced query count, caching, and tuning of underlying queries.

· Integrated and aggregated data from CSVs, Spark outputs, application data, and modelling outputs.

· Diagnosed performance bottlenecks in a high-traffic production MySQL v5.7 RDS instance and partnered with developers to implement fixes that reduced database load by 20%.

· Optimised MySQL server configuration using InnoDB engine analysis, reducing disk reads to one-third of prior levels, increasing write requests before flushing by 2–3x, and improving buffer pool utilisation.

· Built a custom monitoring solution and analysed query logs to identify high-frequency queries, caching opportunities, and indexing gaps, improving query efficiency and reducing aborted connections.

· Planned and executed an automated 60-step migration using dynamic queries and Bash scripts to move the production database from a 4.4 TB to 2.5 TB RDS instance with zero downtime.

· Further reduced database size by at least 250 GB (10%) by removing redundant indexes and correcting oversized field definitions, collations, and non-scalable table designs.

· Built a load-aware online schema change solution using triggers and procedures to alter large production tables with millions to billions of rows with zero downtime.