Software Engineer (Backend)
Backend and platform engineer building AI-driven advertising products, large-scale APIs, and data-heavy systems across AdTech, cloud, and commerce.
Bengaluru, India · M.Tech (Computer Science), IIT Madras
Experience
A few initiatives spanning campaign creation, optimization, ML-driven creative scoring, and safe operations. Impact is described qualitatively on this site.
End-to-end AI-assisted flow so advertisers can stand up campaigns, ad groups, creatives, budgets, and targeting strategy with generated defaults—ready to go live in minutes instead of manual setup.
Team Award for successful delivery.
Recommendations layered onto existing campaigns and ad groups: when to adjust targeting, budget, schedule, bids, and more—aimed at underperforming segments without rebuilding the whole campaign.
Bravo Award (2025) for exceptional design of Unified Recommendations integration with Campaign Manager.
Production ML path built on Airflow: extraction and flattening pipelines into cloud storage, training jobs on prepared data, artifact promotion, and deployment DAGs that roll forward only after validation—plus verification between test and production.
Operational tooling with role-based access to manage campaigns and ad groups safely during incidents—reducing risk of large-scale budget overspend when systems are under stress.
Before that
Harvester and SCH workflows for automatic discovery of entities from log payloads and creation of connector-friendly entities—replacing manual, high-churn setup for teams onboarding applications to OCI Logging Analytics.
APIs so OEM partners can see structured deal information in one place for a given window—instead of scattered terms—making it easier to act on programs spanning OS, Office, and related software and grow order volume.
Early work in compliance—how invalid or abusive use is detected and blocked—followed by the global pricing engine: presentment currencies, withdrawal limits, and impact analysis at massive daily transaction scale.
Get in touch
Send a note below. I don’t list a public email or phone here—LinkedIn works well for anything time-sensitive.