Looks like you have JavaScript disabled. For the full Twine experience, you will need to re-enable it.

I’m a data and platform engineer who enjoys turning complex, messy data into signals people can trust. I’ve spent my career working close to product, analytics, and leadership teams, helping them make better decisions with clear, reliable data. I care a lot about data quality, clarity, and real-world impact. I’m motivated by solving ambiguous problems, building things that last, and seeing my work actually used. I value honest teams, clear ownership, and a culture where people care about doing the right thing, not just shipping fast.…I’m a data and platform engineer who enjoys turning complex, messy data into signals people can trust. I’ve spent my career working close to product, analytics, and leadership teams, helping them make better decisions with clear, reliable data. I care a lot about data quality, clarity, and real-world impact. I’m motivated by solving ambiguous problems, building things that last, and seeing my work actually used. I value honest teams, clear ownership, and a culture where people care about doing the right thing, not just shipping fast.

YanTao Zhang

AI Engineer, Data Scientist, Cloud Developer, +5





I’m a data and platform engineer who enjoys turning complex, messy data into signals people can trust. I’ve spent my career working close to product, analytics, and leadership teams, helping them make better decisions with clear, reliable data. I care a lot about data quality, clarity, and real-world impact. I’m motivated by solving ambiguous problems, building things that last, and seeing my work actually used. I value honest teams, clear ownership, and a culture where people care about doing the right thing, not just shipping fast.…I’m a data and platform engineer who enjoys turning complex, messy data into signals people can trust. I’ve spent my career working close to product, analytics, and leadership teams, helping them make better decisions with clear, reliable data. I care a lot about data quality, clarity, and real-world impact. I’m motivated by solving ambiguous problems, building things that last, and seeing my work actually used. I value honest teams, clear ownership, and a culture where people care about doing the right thing, not just shipping fast.

Available to hire

I’m a data and platform engineer who enjoys turning complex, messy data into signals people can trust. I’ve spent my career working close to product, analytics, and leadership teams, helping them make better decisions with clear, reliable data.

I care a lot about data quality, clarity, and real-world impact. I’m motivated by solving ambiguous problems, building things that last, and seeing my work actually used. I value honest teams, clear ownership, and a culture where people care about doing the right thing, not just shipping fast.

Skills

Experience Level

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Language

English

Fluent

Work Experience

Member of Technical Staff at xAI

October 1, 2025 - Present

Designed a unified training + inference telemetry lakehouse on Databricks + Delta Lake, defining table layouts and partitioning to enable cross-run debugging and reduce regression triage time. Built incremental, replay-safe pipelines in Spark/Spark SQL (idempotency keys, late-event upserts, controlled schema evolution), improving key table freshness from ~half-day cadence to ~1–2 hours. Implemented parameterized Databricks Workflows with retries/backoff, SLA alerting, and guardrailed backfill controls, reducing recurring backfill failures. Standardized evaluation artifact tracking with MLflow (dataset snapshots, metrics, input schema checks). Operationalized telemetry quality + drift monitoring with on-call routing and runbooks, improving MTTD/MTTR. Optimized Spark performance for high-cardinality joins (skew mitigation, join strategy changes, file sizing/compaction).

Staff Software Engineer at Google DeepMind

October 1, 2023 - October 1, 2025

Built a reproducible research dataset lakehouse with dataset contracts (manifests, deterministic builds, validation). Delivered end-to-end ETL/ELT with incremental loads and safe backfills, moving refresh from manual multi-day cycles to scheduled daily/near-daily builds. Led cost/performance optimization for heavy Spark workloads (skew/shuffle mitigation via staged aggregation, repartitioning, caching, storage compaction). Implemented a unified evaluation service with MLflow artifacts and warehouse-exported metrics, accelerating go/no-go decisions. Created production monitoring for training/inference telemetry with OpenTelemetry, dashboards, SLOs, and incident runbooks; delivered near-real-time telemetry ingestion with dedupe keys and replay tooling.

Software Engineer at Cresta

March 1, 2022 - October 1, 2023

Built a conversation intelligence lakehouse for high-volume call/chat events on Databricks + Delta Lake, standardizing ingestion contracts and reducing new source onboarding from weeks to days. Implemented incremental + backfill-safe pipelines (watermarking, late-event handling, deterministic keys with MERGE upserts), improving reliability and reducing reprocessing costs. Established data quality validation + reconciliation checks with automated alerting; tuned Spark jobs (join strategies, skew handling, compaction/OPTIMIZE) to improve a daily curation workflow runtime by ~25–35%. Shipped end-to-end ML deployment workflow using MLflow registry with promotion gates; built batch scoring pipelines with idempotent re-runs and online serving integration. Implemented model/data monitoring (latency, drift, feature health) with dashboards and on-call playbooks.

Staff Software Engineer at Google

March 1, 2016 - March 1, 2022

Built streaming ingestion from event streams into a curated warehouse using Apache Beam/Dataflow and BigQuery; designed maintainable warehouse models with optimized SQL (windows, partition pruning, clustering) to reduce costs and improve dashboard performance. Led a large-scale Spark-based feature engineering platform with deterministic contracts and validation. Implemented near real-time KPI anomaly detection with streaming pipelines and production alerting/runbooks. Led a lakehouse migration to Databricks + Delta Lake, consolidating ETL paths into a governed layer and achieving significant runtime/cost improvements. Standardized batch scoring and model lifecycle with MLflow, and strengthened production operations with monitoring, incident response, and postmortems.

Research Assistant at University of Maryland

May 1, 2015 - December 31, 2015

Built reproducible data preparation pipelines in Python and SQL; implemented baseline ML workflows (feature engineering, cross-validation, metrics reporting) with scikit-learn. Used distributed processing for large transforms and documented runbooks to improve lab collaboration.

Member of Technical Staff at Digital Signal Corporation

June 1, 2014 - August 31, 2014

Automated ingestion and normalization for device telemetry logs; improved daily data availability for analysis. Optimized recurring report performance via indexing and query rewrites. Built Linux automation scripts for scheduled runs and log rotation to improve daily ingestion reliability.

Teaching Assistant at University of Maryland, College Park

September 1, 2013 - May 31, 2014

Automated grading pipeline with deterministic test execution and structured reporting; added guardrails for flaky tests and environment issues; produced staff documentation and templates to improve maintainability.