Seattle, Washington, United States
As a Software Engineer and Founder, I bring a wealth of expertise to the table. My journey has been marked by leading high-performing software engineering teams, building intricate machine learning models and architecting robust serverless solutions within the Native AWS ecosystem. Proficient in a range of programming languages, I have a particular fondness for Go and Rust. Recent explorations have drawn me deep into the captivating realm of Large Language Models (LLMs) and embedding models, where my strong emphasis lies in pushing the boundaries of optimizing multi-model LLMs for extreme low-latency requirements, improving accuracy via cutting-edge fine-tuning techniques and low-level model architecture customization. My design philosophy revolves around a core commitment to maintainability and simplicity. Despite a strong reputation for high performance, the journey always starts with the most basic approach possible before identifying the true bottlenecks that require the most attention. This approach has yielded the deployment of state of the art high-throughput, real-time processing systems seamlessly integrated into serverless infrastructures with the highest availability requirements in the industry. I'm particularly passionate about channeling my skills into startup ventures, relishing the challenge of tackling substantial and impactful problems that transcend specific industry boundaries.
Building SOTA meeting intelligence. Backed by South Park Commons Founder Fellowship '25.
- Architected and engineered an internal event-driven document processing platform to enable proprietary chatbot and machine learning algorithms for document analysis, bolster legal compliance, and consolidate programmatic document APIs. The architecture included extensive use of OpenAI, Anthropic, Mistral and other document embedding models. Core capabilities of this platform powered the new Relay product, enabling secure and scalable customer document processing with embedded ML. - Created a custom machine learning model to extract key terms from semi-structured Excel data such as pro forma documents. Overcame challenges posed by non-standardized, positionally dependent data formats to achieve automated handling and extraction. - Co-led the development of a new mono-repo architecture utilizing Nx and the AWS CDK to facilitate developer self-service application and resource creation. This resulted in over 12 new applications including a new customer offering since the architecture was established. - Engineered a scalable wrapper around 3rd party GPT models that manages outages and scaling constraints for organizational GPT integrations with OpenAI, Anthropic and others. Leveraged advanced serverless patterns to intelligently retry errors and send asynchronous responses within a sub-100ms SLA while supporting ~3000 TPS. - Led the re-architecture of our automated filing services (Bender), taking it from <20% to ~100% filing success rate for EINs, BlueSky, and FormD filings for thousands of funds every month, saving the company >100k legal contractor hours in the process.
- Led the 0 → 1 development of ORSNN’s loan trading platform, building core systems from the ground up. - Designed and implemented Automated Due Diligence using advanced LLM techniques, enabling large-scale, realt-time extraction and verification of loan details against collateral files. - Re-architected core infrastructure, reducing average API latencies from > 10s to < 2s across the application, and migrated the entire stack to AWS CDK for full Infrastructure-as-Code (IaC). - Built and deployed a comprehensive CI/CD pipeline, integrating automated testing and multi-stage deployments to streamline development and production stability. - Led feature development and platform improvements that directly contributed to closing ORSNN’s first loan trades. - Primary languages: Go, Rust, Typescript, and Python. - Key Technologies: LLMs, OpenAI, Anthropic, AWS, CDK, Serverless.
Built and led a team of 16 engineers specialized in rapid systems development. Delivered over 10 large scale systems that have improved Amazon talent retention by leveraging ML, and won the “Just Do It” company wide award for separate products two years in a row. Worked across Amazon to standardize Golang development processes while providing assistance to organizations that were adopting the language.
Significantly built out and re-architectured a confidential project from an acquired startup that scaled to serve an order of magnitude more users while reducing operational costs and adding multiple more engineers to the team. The micro-services were all large serverless AWS systems built using the Golang programming language. In addition to owning several portions of the architecture, I was able to generalize my Golang development process improvements and knowledge to streamline the process across the team and the larger community at Amazon by contributing to multiple large community projects that helped to make one click golang deployments in CI/CD pipelines standard at Amazon.
Designed large scale systems that significantly reduced the product return rate at Amazon in order to improve the Positive Customer Experience in Amazon Global Fulfillment Services. - Used JSP to build the Voice of the Customer page in Seller Central which allows seller's to take preventative action to reduce negative customer experience and to better understand their customers. - Utilized ElasticSearch, DynamoDB, API Gateway, AWS Lambda, and other AWS services to design several backend native AWS systems in order to collect and analyze a high number of transactions per second to determine individual seller's negative customer experience rates. - Improved my team's operational efficiency by designing and implementing full CD pipelines with complete testing. - Introduced Golang onto my team to get lambda coldstarts down from >30s to <1s