Loading public profile
An AI software engineer focused on agent evaluation, model optimization, edge LLM compression, and multimodal machine learning. Experience includes building agent failure-analysis pipelines, MCP-based developer tools, quantized and distilled language models, geometric planning systems, and on-device computer vision applications.
Skills
Experience
Research Assistant, Agent Harness and Model Co-Optimization — Language Technologies Institute, CMU
→
Advised by Prof. Graham Neubig. Owned the data-selection track of a joint harness-and-model optimization project, turning failed agent trajectories into the training-data signal for rejection-sampling fine-tuning (RSFT). Built a failure-analysis pipeline over 720 agent trials of raw JSON telemetry, classifying each trial by exit state and exposing a 15.4% silent failure rate the reported mean absorbed, a 10-point AIME swing. Selected evaluation subsets with nearest-neighbour search over a 1.79M-row dataset, re-embedding all 17,398 DAPO-Math problems with sentence-transformers to rule out a text artifact, with 91 of 100 picks surviving.
Software Engineering Intern — ASUS, Open Cloud Infrastructure Software R&D Team
→
Automated issue triage with a GitLab Model Context Protocol (MCP) agent, pairing an LLM classification prompt with MCP tool calls to auto-classify, label, and route issues, then backfilling the manual backlog. Shipped four context providers into a VS Code AI coding assistant, adding commit-history, OS-metadata, current-file, and repository-map context selected from an evaluation of candidates. Cut a full regression pass from approximately 16 hours to approximately 1–2 hours with a TypeScript end-to-end (E2E) test suite covering approximately 95% of manual cases across Linux, Windows, and macOS.
Research Assistant, Edge LLM Compression — Wireless Ad-Hoc and Sensor Networks Laboratory, NCU
→
Recovered 4-bit accuracy at 2-bit model size (IoU 0.9998, 1.18 GB) by fine-tuning a Llama-3.2 student with GPTQ quantization and LLM-Neo knowledge distillation through a LoRA adapter in PyTorch. Evaluated each configuration against 4-bit and 2-bit quantized baselines, tuning LoRA rank and KD loss weighting to isolate how much accuracy distillation recovered.
Startup Engineer — RoomMuse
→
Caught unbuildable layouts at design time rather than on install day with a six-rule geometric conflict engine that blocks any unit sealing a socket, pipe, or doorway while passing intended connections, validated with 45 tests.
Featured projects
Museum Artifact Recognition App
iOS · Swift · TensorFlow Lite · YOLOv8n · Pix2Pix
Shipped on-device model inference in an iOS app, running the team’s YOLOv8n and Pix2Pix reflection-removal models in Swift with TensorFlow Lite at 94.5% accuracy across 7 classes. Curated the 2,000-image training set and 699-image evaluation set spanning varied angles and lighting, and added reposition guidance so the app recovers from failed recognitions.
Dual-Strategy UAV Path Planning for Building Structure Inspection
C++ · Python · A* · KNN · MST · DFS · 3D Point Clouds
Designed a dual-strategy planner trading layered MST-DFS for structured-space traversal against collision-aware A* replanning, cutting path-generation time over 5× and path length up to 14%. Built the 3D point-cloud pipeline in C++ and Python, slicing raw .pcd scans into layers and chaining per-layer KNN and MST into a DFS traversal, with normal-vector correction for full coverage.
Education
Carnegie Mellon University (CMU), School of Computer Science, Language Technologies Institute — Master of Science, Intelligent Information Systems
National Central University (NCU) — Bachelor of Science, Computer Science and Information Engineering
Last updated
