- The client required data engineering and analytics support to gain insights into their product’s Key Performance Indicators (KPIs).
- They sought periodic capturing of competitors’ KPIs to assess their competitive edge in the market.
- Improving the accuracy of content classification into threat categories was crucial.
- An automated evaluation process was needed to replace manual assessments and maintain data for audits and tracking to retain accreditation certificates.
- Devised an advanced automated pipeline architecture using Apache Airflow, offering high flexibility to adapt to changing KPI capture methods and evaluation processes.
- Reusable design facilitated the evaluation of additional languages like Japanese and French.
- Utilized MLFlow to assess classification KPIs and share the source dataset for further training of core ML models.
- Implemented Google Looker dashboard for performance insights, streamlining data visualization and analysis.
- Enabled in-depth investigations into computer vision and Natural Language Processing (NLP) models.
Performant KPIs over competitor to showcase product value
Captured to generate insights across 8 dashboards
Additional languages supported, enhancing global scope of brand safety efforts