Alex – Python, React, AWS

A large-scale data annotation platform designed to coordinate over 100,000 domain experts in producing high-quality datasets for training and fine-tuning advanced AI models. The system ensures consistency, accuracy, and scalability in handling complex data labeling workflows for enterprise-level AI applications.

• Led a 7-person cross-functional team; established the technical roadmap and agile rituals;
• Introduced a CI/CD pipeline on GitHub Actions and enforced code review standards, improving deployment frequency and code quality;
• Architected the core platform using Django REST Framework;
• Optimized performance by fixing N+1 queries, adding PostgreSQL indexes, and profiling slow queries, cutting p95 API response times by 50% on key endpoints;
• Built a context search feature using text embeddings, enabling experts to find relevant tasks via semantic similarity;
• Developed the main app with React and Redux, optimizing performance with list virtualization and SWR for efficient data fetching and caching;
• Designed a dynamic annotation interface engine that parses XML and form-builder configs at runtime to render custom data-labeling UIs without frontend redeploys.

The core platform and data infrastructure of a scalable headless CMS, supporting thousands of developers and enterprise users.

• Engineered critical components of the OAuth 2.0 authorization server and client flows, enabling secure third-party integrations for Storyblok's marketplace. This involved implementing secure token handling, refresh token rotation, and scope-based permission systems to protect user data;
• Designed and built robust ETL (Extract, Transform, Load) pipelines to ingest and synchronize financial data from Stripe and Revolut APIs into the company's data warehouse;
• Architected an offline conversion tracking system that correlated marketing campaign data with SaaS subscription events. By processing UTM parameters and aligning them with Stripe customer lifecycle events, the system provided data-driven insights into CAC (Customer Acquisition Cost) and LTV (Lifetime Value), enabling the marketing team to identify and double down on high-performing channels and keywords;
• Contributed to the main marketing website and developer documentation, implementing A/B tests and optimizing landing pages for conversion rate optimization (CRO) based on data insights.

The project focused on building a platform to optimize collateral management for large financial institutions. It enabled efficient allocation of cash, securities, and bonds across multiple trading deals, helping to minimize costs and risk while ensuring compliance with regulatory and operational constraints. The system provided analytics and automation tools to support treasury and trading teams in making data-driven decisions.

• Designed and implemented a scalable data pipeline using Python to automate the ingestion and validation of millions of daily transactional records from upstream systems. This replaced significant manual Excel-based processes, reducing operational workload by an estimated 35%;
• Developed ETL processes to cleanse and structure data, feeding it into the team's core C++ optimization models. Gained experience with high-performance computing by assisting in the creation of Python bindings for C++ libraries to expose model functionality;
• Built internal React and Next.js web applications to visualize model outputs, collateral allocations, and margin impacts, providing traders and treasury staff with data-driven insights.

A digital insurance marketplace that enables providers to list products and agents to distribute them.

• Managed a team of 5 engineers, introducing improved development workflows and contributing to technical hiring and mentoring;
• Led the development of the primary GraphQL API using Django and Ariadne, which consolidated multiple potential REST endpoints into a single, flexible interface for the React frontend, improving data fetching efficiency for the client applications;
• Oversaw the development of the customer-facing React/Next.js application. Architected and built a separate real-time chat service using FastAPI and WebSockets to facilitate communication between customers and agents, deployed alongside the main application on AWS ECS.