Michael Kuhn

About Me

🦡 I am a recent graduate from the University of Wisconsin - Madison, where I studied math and computer sciences.

🦡 I am excited by problems that can be tackled using data and computation.

Research Interests

🦡 Geographic data analysis for real-time tracking and routing

🦡 Applying modern ML tools such as neural nets, diffusion, and RAG, to problems involving real-world data. Some problems I’m interested in are image classification, object detection, and forecasting.

Highlighted Projects

Suggesting Routes for Strava Users

Web app to filter community-generated routes based on distance, elevation gain, and location → We find the most relevant routes. Users are able to login and upload data through their Strava account.

Tags: Django, PostgreSQL, Mapbox API, Strava API, OAuth

Code Demo

Streaming Weather Data With Kafka

Follows the producer-consumer model, where the producer is replicated across multiple servers and generates weather data. The consumer continually produces stats through atomic JSON writes and offset checkpointing.

Tags: Kafka, Docker, Python, JSON

Code

Graphics Town - Computer Graphics final course project

3D town showcasing manual vertex manipulation, texture mapping, lighting, and object creation. Built the windmill and fence posts procedurally from Three.JS primitives, implemented all coordinate tranforms for movement, as well as the tractor 3D-movement logic.

Tags: Three.JS, OpenGL, 3D Geometry, Physics

Code Demo

Coursework

(approx. ranked by relevance)

Linear Optimization

Theory behind the simplex method, duality, and solutions to optimization problems constrainted by linear inequalities.

Theory and Design of Programming Languages

Functional, imperative, and object oriented design. Analysis using formal logic (lambda calculus, imperative proofs) and real-world implementations using Haskell, Node.js.
Built a Haskell parser, interpreter for the bf language, and lots of maps/folds/zips/filters.

Big Data Systems

Deploy and use distrubuted systems to store and analyze large datasets
Used Kafka, Spark, Docker, HDFS, BigQuery, Google Cloud, PyTorch

Robotics

ROSpy, sensors, kinematics, motion planning, dynamics and control of serial chain robotic manipulators.
Final project controlling a robotic arm to execute chess moves. Link to project demo.

Large Language Models in Industry

In-context learning, "prompt engineering", RAG, and recent research discussion
Final project using LangChain to analyze accuracy of different document-chunking methods, imposed by constraints of context windows. Link to project.

Machine Learning

Introduction to machine learning for pattern classification, regression analysis, clustering, and dimensionality reduction
Heart disease prediction using SVM/Random Forest for final project presentation

Numerical Analysis

Polynomial forms, difference methods, approximation, splines, orthogonality, solving IVPs/BVPs

Mathematical Methods in Data Science

Matrix factorizations, finite Markov chains, intro to convex optimization

Algorithms

Paradigms for the design and analysis of efficient algorithms: greed, divide-and-conquer, dynamic programming, reductions, and the use of randomness. Computational intractability including typical NP-complete problems and ways to deal with them.

Computer Graphics

Image representation, formation, presentation, composition and manipulation. Modeling, transformation, and display of geometric objects in two and three dimensions. Representation of curves and surfaces. Rendering, animation, multi-media and visualization
"Graphics Town" final project integrating (most all) course themes into a single environment (see highlighted projects!)

Stochastic Processes

Poisson processes, renewal processes, queueing, branching, discrete and continuous-time Markov models