Resume
Skills
Programming: Go, TypeScript, JavaScript, Python, Rust, Elixir, HTML & CSS, SQL
DevOps: AWS, CDK, Docker, Terraform, Git, Datadog
Databases: DynamoDB, PostgreSQL, MySQL, Redis, ElasticSearch
Frameworks: Node.js, React.js, Angular, Phoenix LiveView, Puppeteer, Midscene.js, Ginkgo, FastAPI, TailwindCSS
Other: Distributed Systems, Event-driven Architecture, Serverless, Microservices, REST API, API Design, CI/CD, Testing
Experience
Money Forward, Inc
Software Engineer
May 2025 – Present
Tokyo, Japan
TypeScriptDockerAWSMySQLRedisElasticSearchTerraformDatadog
- Reduced SaaS integration implementation time from 1-2 days to ~30 minutes by building an agentic browser automation system with an OpenAI-powered planner and Midscene.js across 10+ integrations
- Improved worst-case user provisioning latency from 1-2 hours to ~5 minutes by introducing a priority queue to isolate critical operations from nightly batch jobs across thousands of workspaces
- Unblocked a major enterprise customer by designing a tenant-aware OAuth callback proxy on AWS Lambda using state-suffix encoding, avoiding weeks of work re-registering 80+ third-party OAuth apps
- Built 30+ SaaS user management integrations via REST APIs, Puppeteer scrapers, and a custom GitHub App
- Unlocked user management for any customer-managed SaaS application by integrating Microsoft Entra ID as an IdP, eliminating per-app integration work
- Eliminated customer-reported disruption from environments stuck in maintenance mode by implementing an automated retry command on a CloudWatch schedule
Amazon Development Center Germany GmbH - IDEs
Software Development Engineer II
Jun 2021 – Sep 2023
Berlin, Germany
GoTypeScriptDockerCI/CDAWSAPI GatewayCloudTrailCloudWatchDynamoDBEBSEC2ECRECSEFSEventBridgeFargateKMSLambdaS3SQSSSMStepFunctions
- Enabled zero-downtime launch of Amazon CodeCatalyst Dev Environments by leading large-scale load testing and resolving critical bottlenecks across DynamoDB, Step Functions, and Lambda prior to release
- Designed and led implementation of an event-driven resource lifecycle system via AWS EventBridge, eliminating GetMetadata API polling and delivering automated status events to 2000+ users
- Identified a critical cross-host container-isolation vulnerability through threat modeling, where randomized container IDs served as the only security boundary between hosts, enabling potential container compromise on unrelated hosts
- Eliminated recurring microservice crashes (previously every 2-3 days) by using Go's pprof to identify and resolve memory leaks, reducing memory usage by over 90%
- Prevented recurring weekly customer data loss scenarios by implementing a fallback recovery boot path for storage-exhausted MDE Dev Environments, allowing users to retrieve their data
- Reduced HTTP request validation errors by 99% by developing a custom Go-based generic request validator, delivering more precise error messages than standard AWS API Gateway integration
- Accelerated test infrastructure iteration by decoupling E2E tests and customer-behavior-simulating canaries from microservice rollout via a dedicated TypeScript-based CI/CD pipeline
- Led ORR approval for production deployment across 40+ AWS regions by implementing safety protocols, 200+ alarms and dashboards, operational tooling, and documenting 300+ dependency API limits
- Reduced operational response time by building a self-updating multi-region operations dashboard with instant CloudWatch access for any combination of microservice, region, and deployment stage
- Expanded image source flexibility for enterprise customers with internal registries by building a Go-based Docker credentials helper for private AWS ECR images
- Enabled customer Dev Environments to launch with startup scripts requiring API keys and secrets by integrating AWS Secrets Manager to inject them as environment variables at boot time
- Implemented enterprise security and compliance requirements including IAM tag-based access control and full AWS CloudTrail audit log integration for customer resources
- Mentored 2 engineers in load testing methodology, both of whom subsequently designed and executed load tests independently for new service features
Deloitte Consulting GmbH
Consultant
Sep 2019 – Apr 2021
Berlin, Germany
PythonTypeScriptAngularS3Google Maps API
- Reduced development time by ~40 hours per web study by building a full-stack behavioral research platform with an Angular frontend, FastAPI backend, and S3-compatible storage
- Optimized data analysis performance by introducing result caching and improving algorithmic complexity, achieving 90% runtime reduction and saving 2 hours per week
- Secured a follow-on client engagement by analyzing over 1 million ship position samples, building a statistical model to identify waiting segments with Google Maps visualizations
- Enabled more accurate marketing targeting by segmenting 5000+ study participants into 6 behavioral groups via statistical analysis of subconscious preferences using various statistical methods in Python
- Led the requirements definition for a verification tool and monitoring dashboard for deep learning image labeling models, securing a new client engagement
Humboldt University of Berlin
Teaching Assistant
Aug 2016 – Mar 2019
Berlin, Germany
RStatistical ProgrammingMathematical Modeling
- Co-led course of 25+ students on statistical programming with R and assisting with assignments like finding optimal option pricing and simulating cancer growth in patients
- Created sample solutions like spam filter based on Naive Bayes classifier and a tool to denoise images using Nadaraya-Watson estimator in R for statistical programming course
- Supported 200+ Computer Science and Math students over 5 courses on linear algebra and statistics, guiding them through optimal assignment solutions and conducting learning sessions for exam preparation
Education
M.Sc. Mathematics
Humboldt University of Berlin
2016 – 2019
Berlin, Germany
B.Sc. Mathematics
Humboldt University of Berlin
2012 – 2016
Berlin, Germany