AI-augmented cloud platform allowing users to integrate software services in a low-to-no-code workflow automation builder. Automations could be run from the web, through a Slack app integration, and automatically via webhook or event triggers.

• full stack (backend focus) with on-call shifts in a fast-paced startup environment with quickly evolving, sometimes ambiguous product requirements

• distributed system with event-driven architecture, microservices

frequent collaboration, but features personally owned included:

• built feature allowing automations to pause and request input asynchronously, resume once input is provided, and reference the input in further logic

• built notification system around automation lifecycle events

• refactored event-driven architecture using SNS and SQS to fan out events to different queues, decoupling publishers from consumers for performance and extensibility

• redesigned automation data model to be more modular and reusable

• performance improvements such as pagination, partial fetching, caching, batching

MIT M.Eng. thesis research. Built machine learning software in Python using PyTorch, aiming to make power supply circuit simulation faster and more convergent.

Designed and tested a prototype stereo vision system for hopping subterranean robots. Built software in Python to generate a 3D point cloud reconstruction of the environment using video data from a moving stereo camera.