Cuducos

I work on the AT Protocol data layer behind Discover: turning the public stream of records — every post, like, follow — into real-time features ML engineers use, at a scale and speed that keeps recommendations live. - Designed and built the feature store behind Discover and other ML products, letting data scientists express what was impossible before and directly shaping Discover's quality: it stores and serves features from the live stream of records — over 10k feature writes/s, tens of thousands of reads/s, all at millisecond latencies, with a 75% cache hit rate. - Stood up the new Discover engine end to end, 20% higher like-rate than legacy and noticeably faster for users: staging, zero-downtime rollouts, and profiling throughout — serving latency down from seconds to sub-second P99. - Designed and built the seen & served tracker giving the ML team a new layer of explicit signals to rank on: tens of thousands of checks per second in single-digit milliseconds, with over 95% cache hits. - Rebuilt the Kafka to BigQuery analytics pipeline so ML and product teams have high trust in client event data for analysis, UX, and internal decisions: rewrote the writer on BigQuery's Storage Write API — sustained high throughput, analytics and live events always in sync.

It is my responsibility to refactor the backend of the product for scalability. The criteria for success are reducing cloud, increasing code quality and technical culture, as well as allowing the platform to onboard 100 to 1000x more users. • Design and implement modules for reusability in different parts of the stack • Identify performance bottlenecks and implement alternative solutions in code and architecture. • Create patterns and documentation for code quality, as well as automate code quality checks • Promote proactivity, transparency and collaboration in tech conversations within the team

I work at the Quality & Trust Automation building tools to improve the quality of third-party apps from the Shopify App Store. • Implemented merchant-facing features using generative IA related to Shopify's App Store; • Implemented a new payout system for third-party developers using multiple payment methods, such as bank accounts, PayPal, and others; • Automated the queue for manual audits in the Shopify App Store, allowing our operations team to invest time in the most pressing issues; • Enhanced tooling for the operations team to better understand the quality of third-party apps; • Automated dozens of quality checks for every app available in the Shopify App Store; • Prototyped new implementations and solutions using Go and Rust for higher performance and availability; • Contributed to open-source libraries maintained by Shopify.

I worked on three different fronts: operations, corporate, and knowledge sharing. On operations, I contributed to projects in Africa and Asia: Mozambique (digital response to COVID-19 at the level of the Ministry of Health's National Health Institute), Nigeria (connecting the local civic tech communities and civil society engagement to empower monitoring and evaluation platforms for local sub-projects), Kazahstan (theory of change for a project focused on artificial intelligence and smart cities), Guinea (designing the implementation for an integrated system for citizen engagement at three different levels; national, sub-national, and community levels), and Niger (discussing the digital infrastructure and the strategy for the implementation of an early-warning system). Also, I was part of the Citizen Engagement Secretariat, joining the discussion with World Bank staff and other civil society organizations providing feedback for on-going operations in countries such as Tanzania, Iraq, and Malawi. On the corporate side, I delivered, with the Agile team, three workshops in Kazakhstan for the country and regional offices and members of the federal government. There and in the headquarters in DC, I also delivered clinics focused on agile methodologies, mainly Kaban and Scrum. Finally, on knowledge sharing, I co-organized and acted as a panelist in the panel "Government as a platform and Civil Society participation" at the Global Partnership for Social Accountability Forum (2019), contributed with posts to the Governance blog on civic tech and social media accounts, and co-authored a paper on fraud detection in the public sector.

As a developer I engaged in technical challenges like porting large amounts of data from NoSQL to graph architecture, updating internal Python libraries and contributing to the the team with code reviews and technical solutions to production bottlenecks.