• Combined publicly available AI models to create new products and use cases like AI background replacement and product placement.• Added fallback logic between AI backends to increase uptime from an average of 70% to 90%.• Designed an alternative, low-cost solution for context lookup using Elasticsearch for GPT based completions models.• Designed a suite of Nest cache, ratelimit, and circuit-breaker annotation logic for use across all of Jasper's ecosystem.• Created the integration between Jasper and Google Cloud Marketplace to allow GCP customers to purchase Jasper seats and features using GCP credits.

• Designed and maintained an enterprise knowledge graph with support for multi-tenancy, authentication and authorization on EKS.• Designed a stateless Spring Boot API to enable front-end developers to use data in the knowledge graph via REST calls.• Wrote a toolkit to administer multiple database clusters with configuration files and CI/CD.• Developed a scalable, event-driven data ingest service using SQS, Lambdas, and Kubernetes to allowin-house and third-party developers to upload their own data to the graph.• Added full-text search support to the knowledge graph by employing DynamoDB and Elasticsearch as cache layers for node and link lookup.

• Designed and built robust API backends using Node.JS, Postgres and Keycloak.• Made modern SPA frontends using Vue.JS, Koa, and GraphQL for a full suite of NLP analysis apps.• Modernized legacy applications and allowed them to be run in scalable containers on EKS.• Built a SaaS frontend to process customer payments and provide product support for Semiring’s microservices.• Built a browser based graph visualization tool to render thousands of nodes and links in under a second.

• Optimized an in-house application by rewriting code to employ multithreading and parallel processing.• Designed a domain-specific language to asynchronously fetch and transform incoming JSON data.• Developed a library to serialize JSON graphs efficiently, saving time and memory by up to 50%.• Created a system to orchestrate hundreds of AWS EC2 instances at once to handle upwards of 1 million requests per hour.