SDET / QA Platform Engineer / AI Quality Systems
At Binance, I work where QA meets backend systems, crypto workflows, CI/CD, observability, and AI tooling. My best work is not a test case. It is the platform around the test case.
Interview mode
Systems lens: I connect automation frameworks, CI/CD, dashboards, alerts, and AI assistants into one quality operating layer.
Systems map
`mTest` generates evidence. `qa_gateway` turns evidence into dashboards, alerts, workflow triggers, and AI-readable intelligence.
Python/Pytest automation for crypto custody flows: API clients, Selenium admin UI, Appium iOS approvals, E2E transfer journeys, reports, screenshots, and result upload.
Flask QA platform that reads automation results, visualizes trends, triggers workflows, powers QA GPT/RAG, sends WEA alerts, and reviews PRs with SafuGPT.
Career timeline
Built and maintained automation platforms for crypto/blockchain systems, including QaGPT, PR review automation, approval monitoring, E2E web/mobile/API frameworks, CI/CD quality gates, and real-time QA dashboards.
Designed automation architecture and CI/CD for automated retail store deployments, working across Python, Kubernetes, Docker, Selenium, and Azure-backed systems.
Built data processing and simulation pipelines for ranking prediction with Python, MSSQL, Flask, and supervised ML models.
Built REST APIs, monitoring UIs, PixiJS components, and route optimization using Dijkstra's algorithm for industrial and financial systems.
The lab
These are the ideas I keep returning to: evidence, consistency, observability, and automation that can explain itself.
QaGPT turns DB test data and documents into queryable insights, summaries, charts, and alerts. It reduces manual dashboard work and helps engineers ask better questions.
Web admin action, mobile approval, API state, transfer history, and explorer links are validated as one business workflow, not isolated UI checks.
GitHub diff ingestion, hunk-level chunking, SafuGPT file review, PR summary, quality score extraction, and optional auto-approval above threshold.
Publications and patents across ML, SDN, Docker automation, and traffic classification shape how I think about systems and data quality.
Evidence board
Connected Selenium web actions with Appium mobile approvals in the same Pytest flows, then verified final state through transfer history and external links.
Built a central platform for dashboards, workflow triggers, WEA alerts, QA GPT, RAG, PR review, and automation result analytics.
Integrated LLMs, RAG, generated analytics code, image/chart output, and DB context to answer quality questions from real execution data.
Worked on smart approval monitoring and abnormal spender approval alerts, strengthening on-chain security awareness.
Added run IDs, execution time, status upload, dashboard links, screenshots, logs, and app/FE/BE version capture around automation runs.
Published work on news popularity forecasting and dynamic network traffic classification using supervised ML in Docker-based SDN environments.
Formation
Master of Science, Computer Science and Technology. Scholarship, top 5%.
Bachelor of Science, Software Engineering. Chinese Government Scholarship, graduation ceremony speaker.
English expert, Chinese HSK-5, Bengali native, Hindi advanced.