• TL in the Anti Scraping Team, Led an extended team of 15 individuals, overseeing operations and driving successful project deliveries resulting in multiple organizational wins.

• Built robust ML systems meeting and exceeding industry standards for user data protection and user safety per the FTC’s consent order.

• Established safeguards aligned with FTC requirements and evolving regulatory frameworks, ensuring Meta's adherence to industry-standard data protection measures.

• Defined and implemented internal bars for uniform data safety across Meta’s platforms, creating a replicable model for proactive compliance.

• Developed and implemented verifiers to continuously evaluate adherence to safeguards, providing real-time insights and addressing vulnerabilities effectively.

• Received the ‘Privacy Mountain Mover’ award for monumental efforts in building the first safeguard for an ML system at Meta, a recognition awarded to less than 0.01% of the company.

• Enhanced the accuracy of ML models by reducing false positives through improved ground truth measurement systems.

• Built anomaly detection systems to monitor label quality, achieving a 60-70% reduction in false positives and detecting potential SEVs early.

• Impact: Improved label accuracy benefiting 2.5M users while reducing manual oversight by 80%.

Deployed real-time dashboards to monitor model performance and identify training anomalies promptly.

• Partnered with regulatory teams to ensure compliance in labeling practices, setting industry benchmarks.

• Led experiments on feature distribution monitoring techniques, significantly reducing downtime caused by data inconsistencies.

• Spearheaded efforts to identify and mitigate scraping risks across Meta platforms by leading a cross-functional team of security and software engineers.

• Developed advanced detection systems for novel scraping techniques and coordinated with legal and policy teams to update Meta’s Terms of Service.

• Published articles to educate stakeholders on scraping risks and mitigation strategies.

• Impact: Successfully converted 80% of generated leads into actionable cases, reducing unauthorized data access and increasing scraping identification by 60%.

• Established an automated appeals process for users incorrectly disabled by enforcement actions, achieving 84% precision in unblocking benign users.

• Designed and authored the initial proposal and logic for filtering rules, ensuring high decision accuracy and compliance with legal standards.

• Impact: Enabled ~80K users per year to regain platform access, reducing manual oversight of appeals and improving user experience.

• Collaborated with the Instagram Growth Team to define appeal process milestones and ensure scalability across additional platforms.

• Enhanced the automation framework to streamline appeals, reducing the average resolution time by 35%.

• Created a comprehensive reporting system to track appeals outcomes and user satisfaction metrics, aiding continuous improvement.

• One of the two core developers responsible for the complete re-engineering of the server side code for the industry leading Flight On-Time prediction algorithm in modern C++(17) using multithreading constructs and OOP design standards to improve the efficiency of the product to create significant savings in both time and space.

• Setup multiple data wrangling projects which accumulated data from multiple sources to setup feature tables for the ML Projects using C++(17) and Python. Automated all the processes using data pipelines.

• Spearheaded the back-end development for the web-based GUI project, involved in the system architecture specifications to designing and construction of the backend architecture in C.

• Provided back-end support using multi-threading programming API’s in Windows for the company’s future GUI product, increased web running speed up to 10x times and reduced memory usage by 70%. Also achieved high levels of parallelism allowing the multi-functioning product to run seamlessly.

• Experienced with the Full Cycle Machine Learning process - Data collection, Data analysis, Data cleansing, Data modelling & visualization for multiple projects in the company. Projects include Taxi time prediction (On-to-In, Out-to-Off) and Runway Prediction. Algorithms Implemented include XGBoost, Randomforest, DecisionTrees, LightGBM and Neural Networks.

• Integral part of the Flight Predictability and Trajectories Team using machine learning/information theory to estimate the time of arrival of aircrafts & continuous improvement software cycles. Involved research and continuous improvement of complicated mathematical algorithms.