Projects

Case studies

Selected commercial projects — focused on the problem, the solution, and the measurable outcome. Details are kept high-level to respect NDAs.

BlueGrid / Recharge · 2023–2024

AI-Powered Merchant Sentiment Analysis Tool

Problem

The merchant support team was manually sifting through thousands of merchant feedback signals across multiple data sources, with no unified view and no way to detect sentiment trends at scale.

Solution

Built an internal Flask application integrating the Google Gemini API to automatically classify and summarize merchant sentiment. Implemented cross-source data integrity checks and automated categorization pipelines, with a clean internal UI for the support team.

Result

Transformed a manual, error-prone review process into an automated workflow. The team gained actionable insights in hours instead of days, enabling faster and more targeted merchant interventions.

Stack

PythonFlaskGoogle Gemini APISQLAlchemyPostgreSQL
BlueGrid / Recharge · 2022

Automated Codebase Refactor via Custom Python Parsers

Problem

A large-scale structural refactor was estimated to take the entire engineering team several months of tedious, error-prone manual work with high risk of introducing regressions.

Solution

Built custom Python AST-based code parsers and transformation scripts that could automatically traverse, analyze, and rewrite the affected codebase patterns. Developed the tooling iteratively with test coverage to validate each transformation step.

Result

Reduced a multi-month manual effort to a single week. Zero regressions introduced. The approach became a template for subsequent refactor tasks on the same codebase.

Stack

PythonAST parsingCode automationPytest
Factory World Wide · 2018–2021

Real-Time Betting Game Platform

Problem

Multiple betting clients needed custom, high-performance game UIs with real-time state sync, complex animation requirements, and tight latency constraints under high concurrent user load.

Solution

Architected and delivered a series of betting game applications using Nuxt.js on the frontend for SSR performance and Vue.js for reactive real-time UI. Python backend services handled game logic, state management, and integrations with third-party providers.

Result

Delivered several game titles on schedule that became core revenue drivers for clients. The frontend architecture established a reusable pattern adopted across multiple subsequent projects.

Stack

Vue.jsNuxt.jsPythonWebSocketsTypeScript