Welcome to my portfolio! I'm a Computer Science and Data Science student at the University of Wisconsin-Madison, passionate about using technology to solve real-world problems.

I’m a problem solver, hacker, and tech enthusiast, always striving to learn and create. My journey started in 2021 when I wrote my first line of code. Since then, I’ve dived into full-stack development, AI/ML, cloud computing, and blockchain technologies, with a strong focus on scalable infrastructure and MLOps.

• Automating MLOps workflows for deployment, monitoring, and scaling AI systems
• Exploring Network Security, Agentic AI, and Computer Vision through self-study
• Automating processes and solving complex problems in my personal projects

• 🏆 6 hackathon wins, including Second Place at One Trillion Agents hackathon, Best Agent Overall at Sonic DeFAI hackathon.
• ☁️ Hands-on with cloud platforms (AWS, GCP, Kubernetes) and MLOps stacks (MLflow, Kubeflow, KServe, Prometheus, Grafana).
• 🤖 When I'm not coding, I attend tech conference, or experiment with the latest AI models.
• 🌐 I'm passionate about making a positive social impact through technology.

• 2nd Place at One Trillion Agents Hackathon
• Honorable Mention Best Agents Overall at Sonic DEFAI Hackathon
• 2nd PLace at Degen Track at SUI Overflow Hackathon

Feel free to reach out if you'd like to collaborate, chat tech, or swap hackathon stories!

FinBud, Chicago, IL | Oct. 2024 - Present

• Deployed high-performance microservices on Amazon EKS with Istio service mesh and NGINX ingress controllers, minimizing response times by 30% and providing fault-tolerant, scalable services to 10,000+ concurrent users
• Delivered 60% cost reduction while ensuring 99.99% uptime by migrating VietStock services to AWS with Terraform, leveraging a mix of Reserved and On-Demand EC2 instances and centralized ELK Stack monitoring of 5,000+ daily logs
• Reduced peak 5xx/timeout errors by 35% via distributed rate limiting and circuit breakers on AWS ElastiCache for Redis
• Optimized 10+ CI/CD pipelines using Jenkins, Docker containerization, and Helm charts to automate integration and load testing with k6, enabling 50+ monthly releases with 80%+ test coverage and decreasing production incidents by 10%
• Revamped the Nuxt.js Stock Watch UX for 5,000+ users with low-latency updates using a stale-while-revalidate caching strategy and dynamic dependency loading, cutting bundle size by 35% and sustaining 2,000+ RPS

Mobifone (Product Platform Team), Ho Chi Minh, Vietnam | May 2024 - Sept. 2024

• Developed an AI-driven customer assistance system using custom-trained financial LLMs with Hugging Face Transformers, achieving 85% query resolution accuracy and serving 10,000+ customers with automated responses
• Refactored a high-performance lakehouse on AWS S3 with Trino, implementing silver/gold layers under a Lambda architecture, cutting query latency by 40% and boosting throughput by 20% across 1TB+ user interaction data
• Streamlined an automation suite with 5 RESTful API endpoints and orchestrated n8n workflow agents for daily sales reporting and market research, improving product development and order fulfillment efficiency by 50%
• Integrated Redis for conversation state caching and vector search, offloading 55% of queries from the primary database

Statistics & Computer Science Learning Centers, Madison, WI | Jan. 2023 - May 2024

• Enhanced 50+ students' understanding of OOP, Python, and data structures through lab sessions.
• Provided over 100 hours of consulting, improving grades by 20%.

• Spearheaded a hand gesture recognition system for ASL accessibility, averaging 10,000+ predictions/day with sub-100ms latency and 95% accuracy by serving TensorFlow Lite models deployed on auto-scaling KServe infrastructure
• Implemented monitoring and alerting for internal ops with Prometheus metrics collection, triggering Kafka repartitioning to handle 8,000+ RPS spikes in offline analytics streams to the Vertex AI feature store, reducing MTTR by 60%
• Processed 500K+ samples from the feature store via Spark ML pipelines on Kubeflow, reducing deployment time by 50%

• Led a team to build a distributed Agentic AI system for real-time crypto analytics, winning 2nd place at the One Trillion Agents hackathon.
• Built a modular backend of 6 autonomous agents for real-time token market analytics, processing 10,000+ tokens and writing 2,000+ updates/sec to TimescaleDB via Kafka and PyFlink to deliver aggregated insights and price-drop alerts
• Achieved 90% fewer hallucinations and 80% groundedness in a token prediction assistant by applying semantic similarity filtering guardrail (threshold 0.75) over 100K+ documents and automating LLM evaluation with TruLens
• Won Second Place at One Trillion Agents hackathon, Best Agent Overall at Sonic DeFAI hackathon

• Built an agentic RAG system using Snowflake Cortex Search and Autogen agents, improving response groundedness by 80%.
• Reduced model hallucinations by 90% using semantic chunking and continuous evaluation with Trulens.

Data Science for Sustainable Development Club, Madison, WI | Sep. 2023 - May 2024

• Led mentorship for 10+ climate-focused teams, providing deep learning and data visualization support to develop sustainability solutions.

• LinkedIn
• GitHub

Feel free to reach out if you'd like to collaborate or chat about tech!