Farrin portrait
💻

Farrin Sofian

Hey there! 👋 I'm a third-year Ph.D. student in Computer Science at UC Irvine, advised by Prof. Stephan Mandt.

My research focuses on controllable and scalable generative AI across audio, text, image, and video, including diffusion models, reward-guided generation, editable systems, and efficient language-model decoding.

I have publications at ICML, ICLR, and NeurIPS, and have worked on scientific diffusion applications at the Chan Zuckerberg Initiative, large-scale ML systems at Adobe, and Turkey's first bird-song identification mobile application.

📚 Publications

Pre-print

Hierarchical Variational Policies for Reward-Guided Diffusion

K. Pandey*, F. M. Sofian*, J. Groeneveld, F. Draxler, S. Mandt

Pre-print, 2026

Pre-print

Variational Test-time Optimization for Diffusion Synchronization

H. Lee*, F. M. Sofian*, K. Pandey, S. Mandt

Pre-print, 2026

Pre-print

Control-Augmented Autoregressive Diffusion for Data Assimilation

P. Srivastava, F. M. Sofian, F. Immorlano, K. Pandey, S. Mandt

Pre-print, 2026

Conference

Parallel Token Generation for Language Models

F. Draxler, J. Will, F. M. Sofian, T. Karaletsos, S. Mandt

International Conference on Learning Representations (ICLR), 2026

Conference

Variational Control for Guidance in Diffusion Models

K. Pandey*, F. M. Sofian*, F. Draxler, T. Karaletsos, S. Mandt

International Conference on Machine Learning (ICML), 2025

Journal

SonicDiffusion: Audio-Driven Image Generation and Editing with Pretrained Diffusion Models

B. C. Biner, F. M. Sofian, U. B. KarakaÅŸ, D. Ceylan, E. Erdem, et al.

ACM Transactions on Graphics (TOG), 2024

Conference

GECTurk: Grammatical Error Correction and Detection Dataset for Turkish

A. Kara, F. M. Sofian, A. Bond, G. G. Sahin

IJCNLP-AACL, 2023

Experience

Research

  1. Mandt Lab

    Working Under Supervision of Prof. Stephan Mandt on Generative Models.

  2. KUIS AI Lab

    Worked on a variaty of projects such as Bird songs classification, audio re-synthesis, and audio-based image editing under the supervision of Prof. Aykut Erdem, Prof. Erkut Erdem and Dr. Duygu Ceylan from Adobe Research.

    Paper GitHub
  3. GGLab

    Worked on creating the first extensive dataset and training SOTA models and baselines for Grammatical Error Correction in Turkish under the supervision of Prof. Gözde Gül Sahin.

    Paper GitHub

Professional

  1. Machine Learning Engineering Intern at Adobe

    Worked on improving the AI Assistant in Adobe Acrobat.

  2. Machine Learning Engineering Intern at RadiusAI

    Worked on camera calibration based on scene geometry. Co-created the company's first fully automated camera calibration system, significantly reducing the company's reliance on expensive annotators.

  3. Software Engineer at Digitopia

    Designed the company's main product to provide marketing analytics for large enterprises, which included creating a dynamic PDF generator from user dashboard results using JavaScript, Node.js, and Handlebars.

Projects

Latest
Audio-based Image Editing

Audio-based Image Editing

Enhancing image Generation and Editing through audio cues. Discover how audio can influence and control image editing.

HuggingfaceGenAIMultimodalPytorch
GECTurk: Grammatical Error Correction in Turkish

GECTurk: Grammatical Error Correction in Turkish

The first extensive dataset for Grammatical Error Correction in Turkish along with baseline and .

A. Kara, F. M. Sofian, A. Bond, and G. G. Åžahin, GECTurk: Grammatical Error Correction and Detection Dataset for Turkish. 2023.
NLPGECPytorch
Birds of Istanbul App

Birds of Istanbul App

The first bird songs classification iOS app in Turkey developed for ornithologists, bird watcher, or those who are curious and want to explore birds in their surroundings.

iOSAudio ProcessingPytorchTransformers