Working on XStream, a low-latency, Apache Spark / Flink style data processing platform at petabyte/sec scale

Built scalable custom code execution for efficient UDFs, partnering with AI/ML/web search teams

Substantially improved experimentation workflows (query planning, fast resource provisioning, interactive authoring UI, pipeline DSL)

Part of Grammarly's Data Platfrom, focusing on Real-Time product use-cases, Analytics (including Real-Time), Privacy, Performance.

• Enabled secure and privacy-aware use of user data for ML training and A/B tests

• Streamlined data privacy policies via versioned and audited declarative config, enforced by the platform

• Improved team engineering process, achieving predictable shipping cadence and transparency into upcoming deliveries

Managed Backend/Data Engineering team responsible for data management and signal extraction, powering product features.

• Built a team of 8 from scratch: hired, defined a mission and a roadmap, set up engineering processes, overseen 3 promotions

• Launched data-centric product features: Identity Resolution, Champion Detection

• Launched Privacy controls and Data Isolation, unlocking a new major market segment

• Shifted to a dedicated data stack: Kafka, Spark, S3, Airflow

• Established data quality metrics, used them to shape the product roadmap

Contributed to the larger org:

• Implemented Software Design Review processes

• Implemented consistent Hiring Guidelines and Interviewer Training

Managed a 9-person Backend team responsible for launching new products and verticals inside the existing Criteo ecosystem. based on domain-specific ETL pipelines and ML models.

• Led the development of specialized ad targeting for airline and hotel industries based on domain-specific ETL pipelines and ML models. Successfully launched after 9 month meeting initial launch target, achieved ~10% ad performance increase over an existing horizontal solution.

• Led the development of the Omnichannel product, an ad targeting suite for brick-n-mortar stores that allowed offline purchase attribution to online ads. Enabled a new revenue stream for the company.

• Drove technical designs orchestrating dependencies in 5 teams over several time zones.

• Designed and implemented a company-wide approach to sustainable creation of new verticals within Criteo’s existing technical ecosystem paving the way for at least two more new verticals.

• Served as a domain expert on Spark and ETL applications in a 50-people department. Contributed into 10+ technical designs over 2 years

• Achieved consistent team velocity by animating agile practices such as Sprints and Retrospectives

• Elevated the company’s expertise in data processing frameworks by presenting industry whitepapers