hi, i'm brian chen
san gabriel, ca | pittsburgh, pa
machine learningstatistics
stats & ml @ cmu
making software that makes people's lives better.
Timeline
2026
- Building Candid @
The Fluency Project- Built an Export Studio that turns raw lesson footage into social-ready clips with burned-in bilingual captions and overlays
- Built word-level karaoke captions and in-browser + server-side video rendering with FFmpeg
- ML Engineer @
Nalana- Developing an AI Blender agent that turns natural language into editable 3D scene operations
- Built core API and Blender integration for planning, command execution, scene inspection, and iterative repair
- working with
Carnegie Mellon Racing- Working on camera–LiDAR calibration for our driverless race car, using checkerboard detection, robust plane fitting and reprojection optimization
- Published the SMSL paper on Zenodo
- Second stage of the 2024 science fair project, rebuilt as a self-supervised extension of SLfM (Chen et al., ICCV 2023)
- Camera rotation model reaches 0.68° median error on a 50,000-pair simulated dataset
2025
- contributing @
ScottyLabs- Built frontend with React + TanStack Router for CMU Courses to enhance semester planning experience
- Developed Rust backend APIs to improve course and instructor data accuracy
- ML instructor @
Beaver-Edge AI- Designed and taught a 10-week ML course to middle and high school kids
- Guided students to explore PyTorch, NumPy, and Pandas.
- Interned @
Learningly AI- Developed a machine learning platform for education
- Worked on a machine learning model to help students learn
2024
- Started Spatialized Motion Sound Localization (SMSL) as a science fair project
- Won 3rd place @ among 500+ competitors/teams
LA Science Fair - Motivated by helping people with hearing loss locate sounds; trained and evaluated in simulation
- Won 3rd place @
- Developed an Apriltag-assisted field localization and trajectory following system for the robot
- Qualified for
FRC Worlds
- Qualified for
2023
- Developed's computer vision system
FRC Team 1160- Object detection with YOLO + tracking pipeline
- Deployed on Orange Pi edge device
- Teacher assistant @
Chinese Club of San Marino- Taught 100+ students about Chinese culture and language
- Built a Wordle Bot that averages 3.77 guesses across all 2,309 Wordle answers