Experience

Nov 2024 - Jun 2025

Santa Clara, CA

Led workshops on PyTorch, Numpy, Transformers, RNN, LSTMs, and optimization techniques for deep learning
Guided discussion sessions on DeepSeek-V3 and LLM optimization, bridging ML theory and engineering practice

PyTorchPythonDeepSpeedPandasNumpyQuantizationLoRASlurmMLAKV Cache

Feb 2022 - Sep 2024

Taipei, Taiwan

Engineered ElasticSearch indexing architectures, configuring n-gram tokenization and tuning field-level search weights (titles, descriptions) to optimize search relevancy across varied entity types (users, collections, NFTs)
Optimized Redis caching layers and search infrastructure, achieving a 70% latency reduction and ensuring high-throughput data retrieval for downstream applications
Built high-traffic serverless APIs using AWS Lambda, API Gateway & DynamoDB, and designed scalable cloud infrastructure using Docker and Terraform
Accelerated release cycles by replacing manual component validation with a custom CI/CD pipeline on on-premise GitLab, incorporating unit testing, RBAC, and blue-green deployments to mitigate P0 risks
Spearheaded cross-functional risk assessments to identify and mitigate exploitation vulnerabilities in 3rd-party APIs

Node.jsTypeScriptSolidityPythonethers.jsElasticSearchRedisRDBMSMessage QueueDockerAWSTerraformGitlabLinuxEthereum

Jul 2020 - Feb 2022

Taipei, Taiwan

Designed and executed 10+ A/B testing experiments targeting user behavior (e.g., default booking dates), utilizing Google Analytics and Algolia ranking systems; identified 4 successful variants that drove a 5% increase in overall conversion rate
Designed scalable RESTful APIs using Hapi.js and PostgreSQL, integrating OpenAPI (Swagger) specifications to standardize documentation and ensure seamless consumption by frontend clients
Refactored a legacy PHP monolith to a Node.JS microservice architecture, improving reliability and deployment speed
Centralized full-stack monitoring for 8 servers by implementing Datadog log-forwarding and tracing; engineered custom dashboards to track API errors and traffic, reducing incident triage time by 80%
Spearheaded the rescue of a critical 4,000-line legacy service, eliminating 1 hour of daily downtime by restructuring database fetches and implementing safe rollback deployments

Node.jsHapi.jsReact.jsPythonGolangPostgreSQLPHPKubernetesFirebaseNginxAirflowJenkinsDatadogAlgoliaGoogle Analytics

Sep 2019 - Mar 2020

Taipei, Taiwan

Engineered scripts to automate the extraction, consolidation, and transformation of weekly sales data from disparate internal databases, effectively eliminating manual toil and accelerating the reporting cycle
Implemented cross-validation logic to reconcile individual component sales with parent-model metrics, ensuring high data reliability for downstream business analytics
Analyzed complex procurement datasets, evaluating component performance metrics (such as attach rates and historical sales volume) against defined thresholds to identify unprofitable hardware
Partnered cross-functionally with various teams to present findings on anomalous or questionable data, successfully driving data-backed proposals for the early retirement of inefficient computer parts
Consolidated raw team data into comprehensive tracking workflows, surfacing key variances and trends to guide executive decision-making around hardware complexity

Microsoft ExcelMicrosoft Power BIMicrosoft PowerPointVBA