Projects

A selection of publications and projects.

Publications

AstroBridge

UniverseTBD · EleutherAI Summer of Open AI Research · ML4PS Workshop · Sep 2026

Multimodal framework that aligns astronomical images, spectra, and light curves with language models using a literature-assisted captioning pipeline.

Ibrahim Habib*, Kevin Li*, Trung Nguyen*, Gabriel A. Patron*, Gaurav Sekar*, Nolan Koblischke, Matthieu Le Lain, Shashwat Sourav, Aman Kumar, Kartheik G. Iyer, Joshua S. Speagle, Itai Sfaradi, Michael J. Smith, and Ioana Ciucă. The Great Chat in the Sky: Aligning Exotic Astronomical Data to Language with AstroBridge. Accepted at the 9th Workshop on Machine Learning and the Physical Sciences (ML4PS), September 2026. * co-first authors

arXiv Code HF Collection

Projects & Explorations

Between the Fire and the Sun

Neuromatch Academy NeuroAI · Aug 2026

Exploring the alignment of diffusion models with representations in the ventral visual stream during visual perception.

Write-up Code

Diffusion Models Should Just Go for the Data

Interactive notebook · Jun 2026

An interactive breakdown and from-scratch implementation of direct-data prediction (\(x\)-prediction) in denoising generative models based on Li & He (2026).

Notebook

The Variational Autoencoder

Technical blog & PyTorch implementation · Feb 2026

A first-principles derivation of variational autoencoders paired and a from-scratch PyTorch implementation for face generation on CelebAMask-HQ.

Post Code

NeurIPS Ariel Data Challenge 2025

Kaggle · Silver Medal, 48th of 860 teams · Sep 2025

A deep learning pipeline combining transit-curve fitting and stellar features to extract exoplanet atmospheric spectra from noisy simulated data (Silver Medal, 48th/860).

Details Code Write-up

Engineering

  • SlidesGen Sep 2025: LLM-powered academic presentation generator, built during an internship at Saudi Electronic University (now known as Prince Musaid bin Abdulrahman University).
  • SCU GPA Assistant May 2025: parses university transcripts to project achievable GPAs.
  • Parallel PageRank May 2025: parallelized PageRank over Stanford SNAP graphs.
  • Tansik Algorithm Dec 2024: Gale–Shapley-based stable matching of high-school students to university programs, in C++.
  • Dahih Al-Dofaa Aug 2024: on-device AI assistant that answers students’ questions from their slides and textbooks.
  • Codeforces Jun 2023: 400+ competitive programming solutions; reached Expert (max rating 1888).

Archive of older projects and coursework →