Roman Sinkus

• Architected and shipped a voice-to-voice AI sales-training platform powered by a retrieval-augmented LLM grounded in 14+ hours of course content, simulating live practice in pitching, closing, and objection handling.
• Grew the product to 50+ paying users and $5,000 in revenue, owning the full technical roadmap including architecture, infrastructure, and customer-facing iteration.
• Built the full stack: Next.js/React on AWS Amplify, a FastHTML backend on AWS App Runner via ECR with CloudWatch monitoring, plus Ultravox for voice, OpenAI for language modeling, and Supabase for auth and data.

Next.js, React, FastHTML, AWS, Supabase, OpenAI

• Designed and developed a full-stack application to streamline midterm scheduling for the biochemistry and computer science departments.
• Built a force-directed graph algorithm to balance instructor preferences and constraints, reducing exam conflicts.

React, Node.js, Tailwind CSS

• Developed an integrated robotics simulation environment using ROS 2, Gazebo, and MoveIt 2, enabling detection of 100% of robot-robot collisions.
• Automated system setup with Bash scripts driven by user input, reducing configuration time by more than 10x.

ROS 2, Gazebo, MoveIt 2, Bash

• Hardened API security by migrating sensitive GET endpoints to POST, keeping credentials out of URLs and logs.
• Reorganized the regression testing plan by fixing and augmenting 10+ broken end-to-end Selenium tests, saving the development team 15+ hours each two-week sprint.
• Spearheaded the effort to achieve W3C WAI compliance for 100% of React components in a hospital screener application.

React, Selenium

• Decreased video decoding times by 10x by enabling ffmpeg HEVC hardware acceleration using CUDA.
• Extended and maintained Evertz's internal media-processing API in C++, validating requests and responses with Boost test suites.

C++, CUDA, ffmpeg, Boost

• Built and deployed a full-stack course-review and document-sharing platform for UBC students.
• Wrote a Cheerio scraper to ingest UBC course data into Supabase Postgres via Drizzle ORM, with magic-link auth.

Next.js, TypeScript, PostgreSQL, Supabase, Drizzle ORM, Vercel

• Built a real-time study-space availability platform over a hackathon weekend.
• Applied OpenCV to count occupants in live video, giving students real-time feedback on study area capacity.

TypeScript, Python, React, Flask, OpenCV, Livepeer, PyTorch