Description

Staff-level ML/Ads engineer who ships measurable lift. I design and scale ad marketplaces, ranking, and auctions—plus safe, observable LLM features. At Whatnot I set Ads Delivery strategy and launched supply-aware bidding (+20% median ROAS). At ShareChat I led 8 ML engineers and rebuilt auction/pacing (+20% value). At Meta I delivered an explainable bad-ads model adding ~$250M revenue while reducing prevalence. Earlier: DP for ads ranking with FAIR, petabyte-scale Spark/S3 ETL, and a startup (Retention.ai) serving 10M users. Spike → MVP → production with evals, monitoring, and cost control. Stack: Python, PyTorch, FastAPI, ranking/auction design, AWS/GCP

Industry field of expertise

Languages

English
Native or bilingual

Workplace preferences

Remote only

Primarily works remotely

Whatnot Global Limited,
Software Engineer, Ads
August 2024 - Today (1 year and 10 months)
London, UK
○ Setting technical direction and executing complex projects to build out Ads Delivery at Whatnot (social livestream shopping) ○ Laid out a long-term strategy for Ads Auctions at Whatnot ○ Launched phase-1 of the strategy by delivering Supply-Aware-Bidding, resulting in +20% median ROAS across all ad campaigns
Python SQL PostgreSQL Machine learning
Sharechat UK Limited,
Staff ML Engineer
February 2023 - August 2024 (1 year and 6 months)
London, UK
TECH LEAD MANAGER ○ Managing and leading a team of 8 ML engineers to evolve the Ads Auction system at ShareChat ○ Defined a technical vision for building a principled ads auction system with a focus on driving long-term advertiser value ○ Shipped a significant platform and business model change, resulting in a system that properly aligns platform and advertiser incentives, delivers more advertiser value, and prevents overcharging advertisers ○ Overhauled pacing module delivering a 20% increase in ads value driven by both algorithmic and infra improvements
Machine learning Python GO Redis SQL
Meta Platforms Inc,
Sr. ML Engineer
January 2020 - January 2023 (3 years)
London, UK
LEAD OOPS PROGRAM TEAM) ○ Responsible for maintaining and improving the ML-based system that identifies Bad Ads within the Meta Ads Ecosystem and applies penalties to them. ○ Leading a cross-org team of 4 Eng + 1 Program Manager working on this project
Machine learning artificial intelligence Python SQL