Projects
A showcase of AI and data science projects that demonstrate real-world impact and technical excellence across various domains.
AI-Driven Content Analysis & Classification System
Deloitte • 2024
Implemented comprehensive AI system for automated content review, relevance validation, and risk category classification. Built sophisticated models to perform company-specific relevance checks against strict business criteria, moving beyond vendor tags to ensure precise data quality. AI models classify content into standardized categories including financial performance, legal/regulatory issues, and reputational concerns.
Impact:
40% efficiency improvement, 50% reduction in manual review time. AI-driven relevance validation ensures only truly relevant content proceeds to downstream analysis, significantly improving data quality and analytical accuracy.
Tech Stack:
Enterprise Data Pipeline & Processing System
Deloitte • 2024
Designed and implemented comprehensive dual-pipeline architecture for enterprise-scale data processing. Built Data Preparation Pipeline for reliable daily ingestion and Data Enrichment Pipeline featuring advanced data transformation techniques. Implemented sophisticated 'data flattening' process to create unique entity relationships, enabling precise analysis. Comprehensive logging system with detailed metrics tracking ensures full transparency and auditability.
Impact:
46M+ daily news articles processed with 99.7% accuracy. Automated batching of 1,000 articles ensures stable loads. Advanced monitoring system tracks processing time, data quality scores, and volume comparisons for immediate issue detection.
Tech Stack:
Nursing AI Diagnostic System with Human-in-the-Loop
GWU Research Lab • 2023-2024
Led the end-to-end development of an intelligent nursing diagnostic system. I designed and implemented a Retrieval-Augmented Generation (RAG) system that references a knowledge base of 80+ documented nursing scenarios. When new patient data is entered, the system retrieves the top 3 most similar scenarios to inform its diagnostic suggestions.
Impact:
Crucially, I architected a Human-in-the-Loop (HITL) feedback mechanism. Nurses can provide feedback on the AI's suggestions, which is then vectorized and stored in our Deeplake (Vector DB). This creates a self-improving system where accuracy and relevance continuously increase with each interaction.
Tech Stack:
Multi-modal AI for Autism Analysis
GWU Research Lab • 2023-2024
I was responsible for the entire audio processing pipeline. My primary role was to extract and analyze audio from raw video footage, tackling the significant challenge of low-quality audio in Korean. I developed a noise reduction process using spectral subtraction and a filtering logic to isolate the child's voice from background noise and parental speech, significantly improving the quality of data for the model.
Impact:
This work was critical for enabling the analysis of 'in-the-wild' videos, a key goal of our research. By successfully processing the audio data, I helped create a system that provides objective, data-driven insights to support clinicians, making behavioral analysis more efficient and accessible.
Tech Stack:
Private LLM with RAG
Atos Zdata • 2023
Developed generative AI private LLM for automated document processing
Impact:
Automated RFP/RFI/SoW response generation
Tech Stack:
Pepper Robot Navigation with HoloLens
GWU Research Lab • 2023
Developed a multimodal AI system enabling the Pepper robot to navigate autonomously. The robot uses Microsoft HoloLens for real-time environment scanning, obstacle detection, and spatial mapping.
Impact:
This research aimed to give Pepper spatial awareness for free movement in new environments, with future goals of recognizing and remembering individuals. LLM was used for conversational interaction.
Tech Stack:
Pepper Robot AI Integration for Healthcare
GWU Research Lab • 2023
I took over a stalled project that used a traditional NLP model and a Unity 3D avatar. I completely redesigned the system by integrating the GPT API for fluid conversation and OpenAI's Whisper API for robust speech-to-text and text-to-speech capabilities. The virtual avatar was replaced with a physical Pepper robot for tangible user interaction.
Impact:
This overhaul transformed a non-interactive prototype into a successful project. The new system was not only presented at a university poster session but was also significant enough for my supervising professor to present at an academic conference.
Tech Stack:
AI-Driven Defect Detection for Aerospace Composites
Bauman Moscow State Technical University (Bachelor's Thesis) • 2022
Developed a novel methodology combining Finite Element Analysis (FEA) with Machine Learning to predict structural integrity in aerospace components.
Impact:
Achieved 95-97% predictive accuracy by generating a proprietary dataset from scratch via complex ANSYS simulations.
Tech Stack: