I am a software engineer @ Ellipsis Labs, an ex-Googler and JPMorgan alumnus. I graduated from Carnegie Mellon University with a Master of Science degree, and I received my Bachelor's degree from the University of Warwick in the United Kingdom.
Developed and maintained distributed ad-serving backend systems for Campaign Manager 360, an integral component of Google Marketing Platform.
•
Built scalable, reliable systems for ad serving and targeting across multiple channels, ensuring performance-sensitive, low-latency operations using C++, driving significant revenue impact.
•
Built scalable Python-based data pipelines to generate audience insights, utilizing predictive modeling for optimizing creative rotation and frequency in media planning.
•
Deployed real-time machine learning algorithms using TFX to optimize ad bidding strategies that drive performance at scale and maximize client ROI.
•
Enabled high-quality conversion attribution by leveraging Chrome Privacy Sandbox APIs, aligning with industry-leading privacy standards.
Brand Safety Controls - Publisher Safety
•
Built and maintained machine learning models to enhance brand safety for Display Ads publishers and users.
•
Enabled precise filtering of sensitive ads and topic-specific ads through solutions like Sensitive Classifier Filtering and General Category Blocking.
•
Provided publishers and users with robust tools to block ads aligned with their preferences, ensuring a safe and relevant ad experience.
Implemented VaR-based calculators in Athena environment for Interest Rates and FX products using Python; performed implementation tests for model updates and methodology enhancement
•
Reconciled the VaR P&L vectors in MaRIE (Market Risk Infrastructure Enhancement) framework; on-boarded full-revaluation and sensitivity-based VaR models for NA Rates Securities
VaR Impact Analysis and Model Performance Backtesting
•
Investigated the VaR impact of migrating from FX Discount Market Model to Rates Discount Market Model in Athena for Commodities books; implemented missing interfaces in RDMM and FXDMM to standardize the comparing environment
•
Evaluated the performance of VaR-based models by comparing the daily hVaR with clean P&L vectors and counting the number of bandbreaks; investigated the reasons of bandbreaks e.g. times series issue, market volatility etc.
Stressed VaR Aggregation: Aggregated firmwide P&L data for weekly stressed VaR reporting; developed a Python program to automate the process of generating stressed VaR reports
•
Pegged Currency Study: Analyzed the time series of spot price and implied volatility of a basket of pegged currency pairs, such as EUR/CHF and USD/MAD; constructed artificial time series for adjusting interventions or discontinuous re/devaluations
