Tunyi Tan

Senior Software Engineer with about 11 years of experience. At Google, focus stayed on resilient data ingestion, normalization pipelines, and ML-ready pipelines that could serve consuming ML models reliably. TypeScript and SQL were used to shape core platform infrastructure, including authentication, webhooks, and rate limiting.

Strong ownership showed up through shipping systems that handled major global financial aggregators at high throughput. High-agency environment habits included greenfield engineering on new architectures and careful quality checks before production releases. Career impact is a consistent end-to-end data lifecycle pattern that reduces operational toil while improving model inputs.

Senior Software Engineer at Visionary Education Foundation (2020-11 – Present)

Owned platform features for an education-focused data-centric product by shaping end-to-end data lifecycle workflows and production services. Built ingestion and normalization workflows that supported downstream analytics and ML-ready pipelines. Partnered with stakeholders to ensure canonical models stayed consistent as data sources evolved and quality issues surfaced early.

• Architected normalization pipelines for education and engagement events, ensuring consistent canonical models and reducing manual reconciliation across 12 reporting views.
• Built resilient data ingestion using TypeScript services and validated schemas with SQL migrations, improving freshness for downstream ML-ready pipelines.

Senior Software Engineer at Google (2019-05 – 2025-12)

Engineered core platform infrastructure for financial data processing by designing vendor-agnostic abstraction layers for ingestion and downstream consumption. Led architecture for end-to-end data lifecycle, including authentication, webhooks, and rate limiting. Delivered feature extraction pipelines that produced ML-ready inputs for consuming ML models and model outputs used for automated financial insights.

• Architected core platform infrastructure for vendor-agnostic data ingestion, reducing adapter churn and keeping end-to-end data lifecycle consistent across 4 systems.
• Deployed authentication and webhooks services with rate limiting controls, ensuring secure delivery from major global financial aggregators to internal pipelines.
• Automated feature extraction pipelines and standardized canonical models, improving downstream model input quality and cutting data drift fixes by 35%.
• Integrated production workflows that productionize ML models for income categorization and anomaly detection, improving detection coverage from 71% to 89%.
• Shipped SQL migrations and complex Postgres schema design for streaming records, enabling resilient ingestion at scale while simplifying audits across services.
• Scaled processing to ingest 800000 records per day, ensuring ML-ready pipelines stayed stable during peak webhook bursts.

Software Engineer SDE - III at Google (2015-05 – 2019-04)

Developed ingestion and consumption layers for financial analytics by designing data-centric services that ingested banking and tax data. Built SQL-based storage and pipeline steps that normalized inputs into model-ready formats for consuming ML models. Collaborated with scientists and product teams to ensure automated financial insights outputs were reproducible and correct.

• Engineered data ingestion and ingest transformations for banking and tax data, validating tax transcripts and mapping self-employed borrowers into stable income categorization features.
• Developed complex Postgres schema design with SQL migrations, ensuring consuming ML models received ML-ready pipelines inputs for automated financial insights.

Master of Science – UC Irvine (2013-09 – 2015-04)