Jacob Erickson

Gaapio is an AI-powered accounting SaaS platform that automates technical accounting workflows.

• Designed and built an end-to-end system for ingesting, auditing, editing, and version-controlling Excel workbooks through an AI chat interface — including a Univer-based viewer with clickable cell citations, a deterministic check suite with an eval harness and CI lane, and chat-driven surgical value, formula, and structural edits
• Built a byte-reproducible op-log document-versioning engine with version replay, restore-in-place, AI-edit attribution, and multi-level undo, and automated ASC 842 lease accounting from chat-native intake through deterministic schedule generation and disclosure output
• Architected a bidirectional sync engine between Gaapio's file cabinet and external cloud drives (Google Drive, then OneDrive / SharePoint) with webhook and delta-polling change detection, conflict preservation, and production hardening via Cloud Scheduler and OIDC
• Rebuilt file management on a unified, Realtime-backed documents view, and led migration of multi-tenant document storage to per-company GCS buckets with semantic search moved from Vertex AI RAG corpora to Discovery Engine data stores
• Drove an app-wide dark-mode and semantic color-token migration alongside ongoing reliability fixes across memo, contracts, chat, disclosures, retention, and exports

ACMES (Active Cooling for Multispectral Earth Sensors) is a NASA-funded 6U CubeSat mission validating cryogenic cooling and methane-sensing technologies for small satellite Earth observation.

• Re-architected the satellite ground station telemetry system from a monolithic, globally-coupled Python application into a modular, queue-driven pipeline — reducing the codebase from ~20K to ~10K lines while adding 99 automated tests where the original had none
• Achieved 100x+ reduction in per-packet processing latency by replacing Python-level byte iteration with C-level buffer search, eliminating deepcopy on the hot path, and removing a 1-second polling latency floor caused by polled message routing
• Replaced a centralized message router and deep inheritance hierarchy with bounded queue pipelines and composable pipeline stages, eliminating deadlock-prone cross-thread patterns and cutting per-component complexity by an order of magnitude
• Introduced build/start lifecycle separation and dependency injection, enabling a 99-test suite that runs without Docker or hardware dependencies
• Architected a communication bridge between proprietary packet processing software and OpenC3 COSMOS for telecommanding

• Engineered a scalable backend architecture in Express.js with tRPC and input validation using Yup schemas for end-to-end type safety, applying a layered service pattern for clear separation of functionality
• Established a modular component-based design system with Tailwind CSS to standardize UI patterns, improve development speed, and maintain consistent styling
• Built a contract generation pipeline converting campaign data into preformatted PDFs via pdfMake, integrating AWS S3 presigned URLs for secure, temporary uploads/downloads
• Developed a payment system leveraging Stripe APIs (Payment Intents, Webhooks, Subscriptions, and Connect) to automate athlete payouts and manage recurring billing

• Developed comprehensive anti-dumping and CVD analysis system with automated job queuing, real-time status tracking, and email notifications for failed processing jobs
• Architected and developed a comprehensive CRM system with Google Calendar API integration, contact management, and communication tracking, featuring real-time updates, automated email history fetching, and multi-user collaboration capabilities supporting both email and SMS through Twilio

• As a part of this program, I worked with students across different colleges to build and launch a tech startup. Although I've chosen not to further pursue the product we created, I gained invaluable experience working on a real product and moving quickly

• Communicated daily with clients about designing and implementing changes to USU websites
• Trained new users to effectively use our CMS

• Responded to calls and chats from incoming customers and fixed issues in a timely and respectful manner
• Communicated with members of a team to tackle a multitude of unique problems day-to-day
• Relayed problems efficiently between administrators with a focus on user-facing decisions