Mike Tran's Portfolio

Develop a full-stack dashboard for RDF database using Streamlit, Plotly, Cytoscape.js, and Mapbox to provide interactive graph visualizations
Containerized the complete application infrastructure with Docker for streamlined deployment and scalability.
Integrated Large Language Models (LLMs) using LangChain to enable natural language querying of the graph database.
Implemented a Retrieval-Augmented Generation (RAG) pipeline enhanced with vector search capabilities using MongoDB Atlas Vector Search.

Developed a computer vision pipeline to automatically detect corn ears, count individual kernels, and cluster kernels into rows.
Reduced manual labeling time by 90% by building a preprocessing pipeline: initial detection via contour analysis, separation of touching kernels using convexity defects.
Modified and fine-tuned a pre-trained Faster R-CNN model (ResNet-50 backbone), achieving 99% precision and 97% recall on images with up to 800 densely packed corn kernels.
Implemented a custom DBSCAN-based algorithm to cluster kernels into spatial rows for structural analysis.

Analyzed public brand sentiment across multiple social media channels using Brandwatch, compiling key insights for strategic reports
Designed and deployed real-time, interactive dashboards to present sentiment data and campaign perfor- mance to clients.
Crafted stakeholder-facing reports using a storytelling approach to effectively communicate key insights