Tony (Tao) Lin tonylim512@gmail.com · +1(415)319-4535 · San Francisco Bay Area, CA EXPERIENCE ========== Coupang — Staff Software Engineer, Search ML Offline / Infrastructure (May 2022 - Present) - Architected and co-built the Search Feature Backfill Framework, eliminating a 4-month ranking experimentation bottleneck and giving ranking teams a scalable path to generate historical training features across thousands of ML signals. - Established the backfill platform as a shared productivity layer for Search ML: standardized feature onboarding, reduced repeated ranking-engineer effort, and enabled multiple teams to iterate on training data, feature freshness, and model experiments with far shorter feedback loops. - Improved feature backfill and training-data velocity by accelerating feature merging 2x and TFRecord generation 8x-10x, reducing daily processing from 2-3 hours to about 15 minutes and making large-batch model-data refreshes practical for the organization. - Co-led Apache Iceberg integration for ML feature backfills, turning counting-feature workflows into a self-service data lake model and reducing batch turnaround from 1 week to 2 days for downstream feature owners. - Drove training-pipeline reliability and scalability as a lever for organization-level ML metrics, ensuring ranking teams could refresh training windows, unblock feature adoption, and measure model-quality improvements without waiting on bespoke infrastructure work. - Designed an agentic backfill workflow inspired by reusable AI engineering skills: engineers describe a desired feature backfill in natural language, the agent produces a structured backfill profile/config, validates dependencies and data contracts, triggers orchestration, monitors execution, and generates standardized reports with hard/soft failure gates for human review. - Turned backfill operations into a repeatable Understand-Act-Verify lifecycle, reducing expert-only operational load and enabling feature owners to safely launch validation backfills, inspect pipeline health, and communicate readiness across ranking, infra, and ML stakeholders. - Built high-throughput feature and context pipelines for search ranking, personalization, and conversion attribution, grounding model workflows in product, user, and behavioral signals used across Search and Discovery systems. - Migrated major indexing and signal workflows to scalable orchestration, resolving performance bottlenecks across KR/TW search result page and review-indexing pipelines while improving operational visibility for partner teams. - Served as technical reviewer and quality gatekeeper across signals, frontfills, and backfills; raised engineering standards through design review, code review, rollout guidance, and cross-team incident-prevention practices. - Recognized with two consecutive years of Top Tier performance for Staff-level impact across ML infrastructure strategy, execution quality, and cross-organizational technical leadership. - Selected for Coupang's AI Interviewer Cohort, helping design rigorous AI engineering interview challenges and conducting technical screens for core engineering candidates. Wish — Technical Lead Manager, Marketplace Platform (May 2017 - May 2022) - Led a 4-engineer scrum team and architected large marketplace platform initiatives from prototype through production, delivering million-dollar business impact. - Bootstrapped a marketplace scoring system in partnership with the ML team to promote higher-quality products across the platform. - Led key merchant-platform projects during hyper-growth, including logistics claim flows and a rebuilt merchant announcement system. - Architected and revamped data-driven merchant policy engines adaptive to region, logistics type, and brand requirements, earning a CEO Impact Award for contribution to company growth. Zenefits — Senior Software Engineer, Full Stack (Mar 2015 - Apr 2017) - Led and developed core components of Zenefits payroll products for a fast-growing SaaS platform. - Architected and led a production-line upgrade from single employee bank account support to multiple bank accounts, reducing payroll operation costs by 40%. - Designed unified payroll schedule infrastructure used across Zenefits production payroll workflows. Microsoft — Software Engineer, Online Service Infrastructure (Sep 2013 - Feb 2015) - Built and operated deployment infrastructure supporting Microsoft online businesses including Azure, Bing, Office 365, and Xbox Live across 50,000+ machines. - Created a scalable, highly available deployment validation cloud service and shipped weekly infrastructure-pipeline features. - Introduced an inter- and intra-datacenter publish/subscribe model for data deployment pipelines handling roughly 50 TB of rollout data per day. - Troubleshot and hotfixed live-site infrastructure issues during deployment virtualization migration. VMware — Software Engineer, Software-Defined Data Center (Jul 2012 - Aug 2013) - Designed and implemented a Spring-based backend framework to integrate pre-installed application scripts for cloud virtual machines. - Built deployment and management capabilities for thousands of virtual machines across large-scale storage datacenters with high reliability and parallelism. Education — Academic Background (Sep 2004 - Jun 2012) - Master of Science in Information Systems, Carnegie Mellon University (Aug 2010 - Jun 2012). - Master in Electrical and Computer Engineering, Beijing University of Posts and Telecommunications (Sep 2008 - Mar 2011). - Bachelor in Electrical and Computer Engineering, Beijing University of Posts and Telecommunications (Sep 2004 - Jul 2008). SKILLS ====== Languages: Python, Java, C/C++, C#, JavaScript Generative AI & LLM Systems: LLM application architecture, Retrieval-Augmented Generation (RAG), Vector databases, Semantic search, Prompt engineering, Agentic workflows, LLM orchestration, LangChain, LlamaIndex, LLM evaluation, AI reliability and monitoring, Inference cost optimization, Reusable agent skills, Structured workflow profiles, Human-in-the-loop validation gates, Natural-language-to-config workflows, Automated reporting and observability Backend & Web: Distributed systems, Backend architecture, Django, GraphQL, React, Docker, MongoDB Data & ML Infrastructure: Apache Iceberg, Spark, Hive, Hadoop, Presto, Kafka, Airflow, Luigi, Real-time feature pipelines, High-throughput inference support, MLOps for non-deterministic systems, Redshift, Looker, Treasure Data Cloud & Operations: AWS EMR, Prometheus, Grafana, Production reliability, Cost optimization, SLA management, Large-scale system design cv.ha7ch.com/talin