Description: ML engine to personalise digital display ads.

Tech Stack: Python, GCP Composer, MongoDB, BigQuery, GCS, custom ML pipelines, Domino, DV360, CM360.

Contributions: Airflow integration, end-to-end automation, cross-functional campaign orchestration.

Impact: Increased engagement and conversion KPIs by over 5% compared to generic ads.

Description: Identifying long-term valuable customers.

Tech Stack: Python, VertexAI, BigQuery, Domino, SA360.

Contributions: End-to-end automation, model refactoring, VertexAI integration, automated model deployment, drift detection and retraining en masse.

Impact: Improved ROAS by over 50% compared to standard platform optimisation algorithms.

Description: Dashboard and API integration to track and reduce emissions in digital advertising supply chains.

Tech Stack: Python, MongoDB, BigQuery, Scope3 API, Domino, DV360.

Contributions: Built the product from scratch with end-to-end automation and integrated emissions data into programmatic bidding workflows.

Impact: Enabled campaign optimisation for both revenue and sustainability, leading to emissions reduction of over 10%.

Description: Automated reporting pipeline delivering consistent, multi-channel campaign analytics.

Tech Stack: Python, BigQuery, gSuite API, Domino.

Contributions: Led the Python development engineering team, ensuring clean, scalable code.

Impact: Eliminated manual reporting effort, reduced errors, and significantly increased speed of delivery by over 10X.

Description: Integrated LLMs into campaign creative planning workflows, matching ad copy to contextually relevant publications.

Tech Stack: Python, Streamlit, LangChain, RAG, MongoDB, ChatGPT & Gemini APIs.

Contributions: Built the pipeline and optimised LLM integration.

Impact: Streamlined creative planning, improved contextual alignment and boosted campaign effectiveness.

Description: Statistical power analyses to ensure tests in media optimisation were robust enough to detect meaningful effects.

Tech Stack: Python.

Contributions: Designed the analysis and built tools for robust test planning.

Impact: Improved reliability of experimental outcomes, saving resources such as time by over 5X and ensuring data-driven conclusions were valid.

Description: custom Python wrapper around the Google Meridian MMM package to simplify usage and streamline workflows.

Tech Stack: Python.

Contributions: Designed and implemented the wrapper with clean, modular code, reducing complexity in handling model inputs.

Impact: Improved usability, reduced setup time by over 8X and enabled faster experimentation with MMM frameworks across projects.