Selected work

• Worked directly with the CTO and product manager to analyze the existing delivery process and identify opportunities for automation.
• Prototyped and compared a deep agentic LangChain architecture with a Claude Code plugin architecture.
• Presented technical trade-offs and helped select the plugin-based approach.
• Built a Claude Code plugin with specialized skills and sub-agents for project context, planning, implementation, Salesforce Flow generation, test generation, validation, and deployment.
• Built an MCP server connecting Microsoft OAuth with Salesforce OAuth and Metadata APIs.
• Integrated Jira and Confluence so agents could work with project requirements and engineering context.

• Created an end-to-end AI-assisted workflow pulling requirements from Confluence and Jira and producing validated and deployable Salesforce Flows.
• Supported both greenfield Salesforce implementations and modifications to existing customer environments.
• Provided secure access to customer Salesforce environments through OAuth-based integrations.

• TypeScript
• LangChain
• Next.js
• React
• Tailwind CSS
• Deep Agents
• jsforce
• LiteLLM
• Monaco Editor
• NextAuth
• Prisma
• TanStack Query
• Zod
• OpenAI
• Claude Code
• Confluence
• Jira
• Langfuse
• MCP
• Microsoft OAuth
• MinIO
• PostgreSQL
• Salesforce Metadata API
• Salesforce OAuth

• Built application services and APIs in TypeScript.
• Designed prompt and context engineering for production LLM workflows.
• Implemented RAG over proprietary customer datasets.
• Built structured extraction workflows using LLMs and LangChain.
• Expanded the data platform using Databricks and Azure AI Search.
• Created new data models, datasets, and search indexes for AI workflows.
• Implemented production prompt versioning and safe prompt releases.
• Introduced Langfuse prompt management, tracing, graph-level metrics, monitoring, and fault detection.
• Established integration testing across AI use cases.
• Contributed to the customer-facing React application.

• Took the system from development into production.
• Delivered the functionality as an advanced paid capability of the Expana platform.
• Automated significant portions of product cost-model construction.
• Established production-grade testing, observability, and prompt-release practices for the AI system.

• TypeScript
• LangChain
• LangGraph
• Next.js
• React
• Tailwind CSS
• BullMQ
• Drizzle
• Highcharts
• Prisma
• TanStack Query
• Zod
• Auth0
• Azure AI Search
• Databricks
• OpenAI
• Langfuse
• PostgreSQL
• RAG
• Redis

• Led the high-level design and architecture of the new product.
• Worked with three product and project managers to turn an ambiguous product concept into concrete technical requirements.
• Identified edge cases and system behaviors missing from the original concept.
• Drove architecture reviews with partner teams, senior and principal engineers across S3.
• Led a team of 4-5 engineers and decomposed the architecture into independent subsystems.
• Mentored engineers and guided them through individual component and data-model designs and implementations.

• Decomposed the system into clearly defined subsystems with explicit interfaces and ownership boundaries.
• Enabled multiple engineers to independently design and implement individual subsystems.
• Established reusable engineering practices for distributed dataset design.

The system was essentially implemented. A product decision then paused the initiative before public launch.

• Java
• Kotlin
• Amazon EC2
• Amazon EMR
• Amazon S3
• AWS CDK
• AWS CloudFormation
• Apache Spark
• CI/CD
• DevOps
• Integration Testing
• Unit Testing

• Helped a new Berlin engineering team take technical and operational ownership of an S3 product from the original Seattle team.
• Improved operational stability and DevOps maturity during the ownership transition.
• Identified performance bottlenecks across large-scale datasets and processing pipelines.
• Diagnosed architectural and configuration limits preventing EMR clusters from scaling efficiently.
• Removed resource contention and processing bottlenecks visible only at large cluster sizes.
• Designed and implemented a dynamic Spark configuration engine that tuned runtime parameters according to dataset size, workload characteristics, and cluster scale.

• Reduced runtime of critical workloads by up to 50%.
• Achieved a corresponding reduction in compute cost.
• Enabled workloads to scale effectively beyond 1,000 vCPUs.
• Reduced manual Spark configuration and made execution more predictable across differently sized workloads.

• Java
• Kotlin
• Amazon CloudWatch
• Amazon DynamoDB
• Amazon EC2
• Amazon EMR
• Amazon S3
• AWS CDK
• AWS CloudFormation
• Apache Spark
• CI/CD
• DevOps
• Hadoop
• Integration Testing
• Microservices
• Unit Testing
• YARN

• Built a data pipeline that ingested products from Amazon's retail catalog.
• Implemented multiple enrichment and classification stages.
• Developed automated logic for determining whether products were appropriate gifts.
• Filtered unsuitable, offensive, sensitive, and low-relevance products.
• Built additional classification stages that grouped products into customer personas based on demographic and interest characteristics.
• Ran a successful A/B test comparing automated recommendations with manually curated recommendations.

• The experiment significantly outperformed the manually curated experience, increasing conversion and product purchases.
• Rolled the ML-based curation system out across multiple Amazon marketplaces.
• Reduced dependency on manual merchandising and automated the curation process, creating a scalable mechanism for continuously discovering relevant products from Amazon's changing catalog.

• Java
• Kotlin
• TypeScript
• Spring
• Spark MLlib
• Amazon EMR
• Amazon S3
• AWS CDK
• A/B Testing
• Apache Spark
• CI/CD
• Data Warehousing
• DevOps
• Elasticsearch
• Integration Testing
• Kibana
• Load Testing
• Penetration Testing
• Unit Testing

• Reviewed the architecture and operational model to identify reliability and scalability bottlenecks.
• Delivered architectural and performance improvements required for peak traffic.
• Prepared the system for high-profile retail experiences including Oprah's Favorite Things.
• Introduced integration tests and automated performance tests in CI/CD, end-to-end UI tests, continuous production canaries, and controlled fault-injection and recovery tests.
• Strengthened observability, release confidence, and failure recovery.
• Established stronger DevOps and on-call practices and trained engineers joining the on-call support rotation.
• Led a complete backend and frontend redesign of the internal administration portal.

• Scaled the platform beyond 1,000 transactions per second.
• Improved production reliability and availability.
• Detected scalability and latency regressions before production through automated performance testing.
• Improved recovery behavior under dependency failures.
• Removed scalability and operational bottlenecks from the administration platform.

• Java
• Kotlin
• TypeScript
• React
• Spring
• Cypress
• Amazon CloudWatch
• Amazon DynamoDB
• Amazon EMR
• Amazon S3
• Apache Spark
• Canary Testing
• CI/CD
• Elasticsearch
• Fault Injection
• Integration Testing
• Kibana
• Load Testing
• Unit Testing

• Helped launch a newly formed engineering team for the product.
• Worked with the Product Manager and Engineering Manager to define the initial product direction.
• Built a functional interactive reporting prototype within two weeks.
• Presented the prototype in Seattle to customers, account managers, partner teams, and senior leadership.
• Led technical portions of customer and stakeholder discussions.
• Helped secure approval and funding for production development.
• Worked in an eight-person engineering team to productionize the prototype.
• Defined low-level architecture for individual platform components.
• Built interactive drill-down analytics across campaigns, audiences, advertised products, and other dimensions.
• Helped build an offline analytics system that detected performance gaps and generated optimization recommendations.
• Extended the system with opt-in automation capable of applying selected recommendations automatically.

• Prototype approved and funded for production development.
• Initial production system delivered within a few months.
• Established a delivery cadence of approximately one new capability every two weeks.
• Replaced static reporting with continuously available interactive campaign analytics.
• Provided earlier detection of underperforming campaigns.
• Moved the product from passive reporting toward automated campaign optimization.

• Java
• Kotlin
• React
• Spring
• Amazon DynamoDB
• Amazon S3
• AWS CDK
• AWS Lambda
• Apache Spark
• CI/CD
• DevOps
• Elasticsearch
• Integration Testing
• Load Testing
• Microservices
• Penetration Testing
• REST APIs
• Unit Testing

• Joined the initial three-person engineering team, which later grew to eight.
• Helped take the platform from initial development to production launch within months.
• Owned much of the integration with an external website-builder provider offered to domain customers.
• Designed the integration architecture and deployment strategy.
• Developed APIs connecting the registry platform with the external service.
• Built substantial supporting AWS infrastructure using Infrastructure as Code.
• Performed load and performance testing.
• Led security reviews and penetration-testing activities.
• Coordinated remediation of security findings before launch.
• Worked with external vendors and internal partner teams to define integration contracts and resolve technical dependencies.

• Successfully launched registry.amazon.
• Supported the launch of .moi and .bot Top-Level Domains with external website-builder integration.
• Delivered repeatable automated infrastructure deployments.
• Validated scalability and security before production launch.

• Java
• JavaScript
• Python
• Spring
• Amazon CloudFront
• Amazon DynamoDB
• Amazon EC2
• Amazon Kinesis
• Amazon Kinesis Data Firehose
• Amazon Route 53
• Amazon S3
• Amazon SNS
• Amazon SQS
• AWS CloudDeploy
• AWS CloudFormation
• AWS Lambda
• CI/CD
• DevOps
• Integration Testing
• Load Testing
• Microservices
• Penetration Testing
• REST APIs
• Unit Testing