Skip to content

Projects

Engineering evidence, before decoration.

Software Engineer & Co-Developer: Material Ingestion & Question-Family Backend

PhaseForge

Co-developing the material-ingestion and question-family/verification backend for an AI assessment platform that turns instructor course materials into grounded, exam-ready questions with answer keys.

  • Implemented material ingestion with retrieval-augmented generation.
  • Built question-family, variant, and verification workflows.
  • Contributed schema design, API contracts, and structured-output validation.
Next.jsFastAPISupabasePostgreSQLpgvectorPydantic
Read case studyView GitHub

Independent Research Project · September 2026

SmoothPose

Built an end-to-end 6-DoF object-pose tracking pipeline for four rigid objects across 1,404 frames using monocular point clouds, segmentation masks, camera intrinsics, and template meshes.

Mesh-rendered pose overlay · Suitcase
  • Implemented and evaluated six geometric and temporal approaches, including 24-start ICP, sequential tracking, quaternion smoothing, dynamic programming, and joint optimization.
  • Handled occlusion, symmetry, and reconstruction outliers across the tracking pipeline.
  • Achieved 0.693 mean silhouette IoU and 3.12 cm mean Chamfer distance with no rotation jumps above 60°.
  • Developed mesh-rendered overlay videos and per-frame evaluation tools with Python, SciPy, Open3D, and OpenCV.
PythonSciPyOpen3DOpenCVICP
Read case study

Backend Developer

HornDeals

Built authentication, UT-student verification, and authorization services for a marketplace, including an AWS RDS relational database schema.

  • Implemented authentication and authorization flows.
  • Built UT-student verification services.
  • Designed the relational database schema on AWS RDS.
DjangoAWS RDSPostgreSQL
View React clientView Django client

3D reconstruction and rendering pipeline

Novel View Synthesis

Built an end-to-end 3D Gaussian Splatting pipeline that reconstructed a scene from 99 images, optimized 392k Gaussian primitives, and rendered RGB and depth views along interpolated camera trajectories.

Novel-view trajectory render
  • Extracted video frames and reconstructed camera poses with COLMAP.
  • Trained a 3D Gaussian Splatting scene representation.
  • Rendered aligned RGB and depth views along interpolated camera paths.
COLMAP3D Gaussian SplattingComputer Vision
Read case studyView GitHub

ML Developer — document-understanding pipeline

CV Revive

Built a layout-aware resume-field extraction pipeline using Tesseract OCR, token normalization, bounding-box extraction, and LayoutLMv3.

  • Created an IOB-tagged dataset of 30-plus resumes and approximately 20k OCR tokens for token classification.
  • Improved token-level F1 from 0.0013 to 0.4885 on a dataset of 30-plus resumes.
  • The result is not production-grade: the dataset is small and the label space is large.
Tesseract OCRLayoutLMv3Document AI
View GitHub

Experience and education

Systems work, in and beyond the classroom.

Experience · Spring 2026

Undergraduate Course Assistant

Mentored students in SQL, PostgreSQL, and relational database design while evaluating coursework for more than 200 students.

Education · Expected December 2027

The University of Texas at Austin

B.S. Computer Science and B.S. Statistics and Data Science.

Contact

Let's talk about the work.

Reach out by email for opportunities, technical conversations, or a closer look at a project.

Email Me