I am a graduate student with an academic focus on embedded systems, low-level programming, and hardware-software integration. Outside of coursework, I spend most of my time building compiler and systems project, with a particular interest in performance optimization and machine learning systems.
🔢 Tensor Compiler, this is my implementation of a tensor compiler built with MLIR. It supports a subset of ONNX operations and implements a lowering pipeline from an ONNX graph through MLIR to LLVM IR, with execution via JIT compilation.
- C++ (MLIR/LLVM), Python, Valgrind, Lit, FileCheck
🛍️ Fashion Recommender System, a system that detects fashion objects and learns image representations through Neural Networks. It also utilizes ANN similarity search index to retrieve and rank these image representations.
- Models trained: YOLOv5, Convolutional Autoencoder
- Vector Store: FAISS Flat Index
- Object Detection Model Repo, the object detection model was used to detect fashion objects and obtain bounding box coordinates
- Embedding Model Repo, the model then obtain a 512-dimensional embedding of each detected fashion objects with an embedding model, and utilized an ANN algorithm to find the top-k similar images based on their vector embeddings
📊 A web application that scrapes data from a public internet archive of software engineering and data science articles, analyzes and visualizes the results of the data, including NLP and data analysis.
- Automated Web Scraper: Airflow, RabbitMQ, PostgreSQL, Python, Scrapy, deployed on AWS
- Interactive NLP Dashboard:
- Tech Stack: Python, Django, HTML/CSS, JavaScript, PostgreSQL, Docker, deployed on AWS
- The system fetches data through RESTful API from the database, analyzes the data with data analysis and NLP libraries (NumPy, Pandas, NLTK, etc.), and visualizes the dashboard on the frontend with Chart.js
Languages: C++, Python, C, Verilog, VHDL
Frameworks: MLIR, LLVM, PyTorch, Docker
Tools: Valgrind, Lit, FileCheck, Git, AWS