Pradyut Nair

Data Analyst Intern

TomTom

2024 August - Present

Responsible for analytics and reporting for TomTom's Horizon and Data Analytics teams.

• Built a multi-agent RAG pipeline to review 40+ telemetry events across the organisation, cutting manual review time by ~12h per week and adopted by 3 business units.
• Standardised metric definitions and automated reporting pipelines across 5+ workflows, giving 50+ stakeholders a single source of truth for product performance.
• Designed and deployed 7+ dashboards to track traffic decoding success rates, map availability, and connection failures, enabling engineers, PMs, and senior leadership to identify issues significantly faster.

Data Analyst Intern

Tesla

2023 February - 2024 March

Solely responsible for data analytics and reporting within Tesla's Residential Energy Team in EMEA.

• Automated SQL pipelines, reducing manual reporting by 85% and generating insights for 20+ product and customer reports.
• Developed discrepancy reports, saving the sales team 8h weekly and improving order tracking accuracy by 35%.
• Built Python and PowerBI dashboards for installation issues and fleet metrics, enhancing performance tracking across 3 regions.
• Automated correction of 20,000+ customer records, achieving 100% customer onboarding accuracy for product launches.

• Implemented DiffEdit style semantic masking into Alibaba's Wan 2.1 model to enable video personalisation without training.
• Ran evaluations on the OpenS2V-Nexus benchmark to measure video quality, identity preservation, and motion smoothness.
• Resulting model ranked 2nd (Single-Domain) and 4th (Human-Domain) on the OpenS2V Leaderboard.

• Integrated semantic features using pre-trained vision models (CLIP, ConvNeXt) into SPAI's spectral-based detector for synthetic image detection.
• Developed and tested late fusion and cross-attention strategies to align semantic and spectral representations.
• Improved mean AUC by +4.4% (92.7% overall AUC), improving robustness to real-world image distortions.