Zavier Rahmansyah

Work Experience

N + 1 Research Institute, College of Computing & AI, University of Wisconsin-Madison

Summer of AI Lab (SAIL) Undergraduate Research Intern

Working on AI-related project. My team and I are currently developing a fintech application that utilizes AI to provide personalized financial advice and investment strategies. The project aims to leverage machine learning algorithms to analyze user data and market trends, offering tailored recommendations for financial planning and investment decisions. My responsibilities throughout this summer include:

Developing data extraction and preprocessing scripts from SEC API using Python.
Providing authentication layer using Node.js.
Initializing deployment processes for the fintech application using Docker.
Conducting research on RAG architectures by comparing different approaches; Naive RAG, Advanced RAG, Modular RAG, GraphRAG, and Agentic RAG.
Conducting research to analyze benefits of the application compared to traditional analytic tools.

International Sibling Society

Data Analyst Intern

Currently working as an undergraduate intern for a non-profit organization focusing on sustainable development goals (SDGs). Responsibilities throughout this summer include:

Created data visualizations and regression analysis using Python, Power BI, and Claude AI Tools.
Conducted SDG research and KPI analysis to support strategic decision-making.
Contributed to weekly discussions and strategy sessions.
Monitored platform performance metrics & provide suggestions for future platform.
Collaborated with module leads to set data-driven goals and track progress.
Prepared data visualizations and presentations for general meetings and reports.

Facilities Planning & Management Division, University of Wisconsin-Madison

Student Help at Human Resources Department

Worked as a student assistant. Tasks include:

Filing Employee-Related paper documents
Filing Employee-Related documents using perceptive content and workday
Managing the Front Desk
Organizing & managing files

Der Rathskeller, Wisconsin Union

Restaurant Team Member

Worked as a team member. Tasks include:

Working in a restaurant team, providing food & beverages, and excellent customer services

Projects

MATE - Multilingual Agent Tactical Explainer

Built a multilingual agent tactical explainer for the IBM June AI Builders Challenge. Inspired by the 2026 World Cup, the project aims to bridge the gap between existing commentary and the need for multilingual explanations of tactical decisions in football matches. The agent also include a small vision model that can detect small moments. The challenge theme addressed are:

Understanding & Explanation — AI match explainers, tactical shift analysis.
Fan & Learning Experiences — personalized multilingual World Cup companion, "teach me the game" assistant.

The tools, frameworks, and technologies used in this project include: IBM Granite, Langflow, Docling, Python, HTML, CSS, FastAPI, PyTorch, and YOLO data.

Race-Engineer-AI-Assistant, IBM AI Builders Challenge

Developed an AI-powered solution that improves team racing for the IBM AI Builders Challenge. The focus area I chose is: Performance Insights. The platform aims to help race engineers process information and provide insights to make key decisions during a race. The tools, frameworks, and technologies used in this project include: IBM Granite, Langflow, Python, React, Node.js, FastAPI, and FastF1 data.

Terminal-based AI Agent, Hackerrank Hackathon Project

Developed a terminal-based AI agent for a coding competition. The project was built using Python and Claude API, and it was designed to assist users if they have any questions related to using Claude, Visa, or Hackerrank. The tools, frameworks, and technologies used in this project include: Python and Claude API. Specifically, the program was made using the principles of TF-IDF, not collecting user's data, and providing accurate information to users. The AI model used for the project is Claude sonnet 4.6.

House of Hope Website Redesign, Badger for Good Designathon

Redesigning a hard-to-navigate website for a non-profit organization for a UX Design Competition. Following the competition, I developed a new website for the organization. The tools and technologies used in this project include: Figma, HTML, CSS, and JavaScript.

Hyperstition for Good Writing Competition

Developed a paper for the Hyperstition for Good Writing Competition. The idea behind hyperstition is the research that suggest that writing about AI reasoning with compassion can actually shape how future AI behaves. In other words, we can help influence how AI treats life.

Proximity Match, Claude Builder Club (CBC) Anthropic x College of Computing and Artificial Intelligence

Building an application to help students find their best match for friends, project partners, and other academic collaborations. The project was made using the idea that students may feel uncomfortable to initiate conversations with strangers, but they may be more comfortable to start conversations if they know that they have something in common and if they know they are located within a reasonable distance. The tools, frameworks, and technologies used in this project include: Python, HTML, Javascript, Shell, Vercel, and Claude.

Synchronized Applications Recommender Knowledgebases, Codex Creator Challenge

Recognized as one of the top projects in the Handshake x OpenAI Codex Creator Challenge. I Developed an application intended to address the problem when people don't know what they want for the Handshake x OpenAI Codex Creator ChallengeThe tools, frameworks, and technologies used in this project include: OpenAI Codex, OpenAI API, React, Node.js, Express.js, Figma, Canva, and Model Context Protocol (MCP).

Personal AI Agent: Daily Quest AI Agent

Developed a Personal AI Chatbot Agent using Langchain, React framework for the Single Page Application (SPA), node.js, Express.js, MongoDB for storing data, and FastAPI, Langchain, and OpenAI for the LLM implementation. The tools, frameworks, and technologies used in this project include: OpenAI API, Langchain, React, Node.js, Express.js, MongoDB, and python FastAPI

Personal AI Agent: Morning Briefing AI Agent

Developed a Personal AI Agent to brief my morning with motivation and list of tasks. While I could have just used google calendar and weather app, I wanted to create an AI agent that can integrate information from multiple sources and provide a concise briefing. The project is also beneficial for me to learn how to build an AI agent and how to integrate multiple APIs together. The tools, frameworks, and technologies used in this project include: OpenAI API, Model Context Protocol (MCP) framework, Google Calendar API, and Weather API.

Badger Solar Racing: Race Strategy Team

Working in a subteam of badger solar racing, the goal of the team is to optimize the usage of car's battery to maximize the car's potential. I worked on: Implementing Isolation Forest algorithm to detect outliers in the car's sensor data.

UW-Madison Deliberation Dinners Project

An initiative that brings together UW-Madison undergraduates with diverse ideological perspectives to deliberate on important political issues.

I found this experience to be useful and exciting because not only we get to talk about wide variety of political issues. It also changed my view on how I approach things, meaning to not only standing between "yes" or "no situation", but to actually have a clear standing and idea on where I want to be in this topic. This experience also taught me how I can improve quality of discussions by bringing up one topic over another and how we can further develop a discussion topic when everyone seems to be in a concensus.

Personal Website

This website is built using HTML and CSS, alongside Visual Studio Code, Github, and Netlify. It showcases my background, favorite quotes, projects, work experience, and extracurricular activities.

Project Management: Your Best Choice University Using Multi-Attribute Decision Process

Developed a comprehensive project management paper for selecting the best university based on multiple attributes and decision-making criteria. Worked under Dr. Paul D Giammalvo