Skip to content
View khoda81's full-sized avatar
19 followers · 19 following

Achievements

Achievement: QuickdrawAchievement: Pull Sharkx2Achievement: Pair Extraordinaire

Achievements

Achievement: QuickdrawAchievement: Pull Sharkx2Achievement: Pair Extraordinaire

Block or report khoda81

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
khoda81/README.md

Hi, I'm Mahdi Khodabandeh

ML / systems engineer working across Python, Rust, PyTorch, and TypeScript.

I like building things where modeling, algorithms, and systems implementation meet: compression, search, probabilistic interfaces, performance-conscious data structures, and interactive tools.

Portfolio: https://khoda81.github.io/

Featured work

  • Chronickle — interactive market-intelligence timeline that aligns price movement and real-world news on one scrubbable canvas. Live demo
  • natwalk — navigate autoregressive model distributions in information space, using exact arithmetic-code actions and Dijkstra search in surprisal.
  • mercy — experimental Rust interface for composable arithmetic coding built around a fixed 64-byte boundary representation.
  • logvec — cache-friendly Bentley–Saxe logarithmic data structure in Rust, with benchmark comparisons against BTreeSet and sorted vectors.
  • rust-2048-solver — performance-focused 2048 search in Rust with compact board operations, heuristic evaluation, caching, and benchmarks.
  • dethcod — research project exploring learned lossless compression with transformer models.

Interests

  • language models and probabilistic interfaces
  • neural and classical data compression
  • reinforcement learning, search, and online adaptation
  • Rust / Python interoperability
  • performance-conscious data structures and developer tools
  • interactive, data-intensive applications

Toolbox

Languages: Python, Rust, TypeScript, JavaScript
ML/data: PyTorch, Transformers, reinforcement learning, NLP
Systems/web: Linux, WebAssembly, SolidJS, Vite, Bun
Development: Git, Jujutsu, testing, benchmarking, package publishing

I'm open to part-time remote contract and research-oriented engineering work involving ML, Python, Rust, or systems-heavy AI tooling.

Pinned Loading

  1. chronickle chronickle Public

    An interactive browser application that places financial-market movements and real-world news on a shared, scrubbable timeline.

    TypeScript 1 1

  2. arithmetify arithmetify Public

    An easy to use implementation of arithmetic coding

    Rust 2

  3. arithmetipy arithmetipy Public

    Fast streaming arithmetic coding for Python, backed by Rust and PyO3.

    Rust 1

  4. dethcod dethcod Public

    Research experiments on learned lossless compression with transformer token and vector bottlenecks.

    Jupyter Notebook

  5. basers basers Public

    Blazingly fast, base converter for exact rational numbers, with automatic detection of terminating and repeating expansions.

    Rust

  6. zoorkhaneh zoorkhaneh Public

    Experimental Python toolkit for reinforcement-learning agents, training loops, and Gymnasium experiments.

    Python 2