Resume/CV

Koray Ulusan Machine Learning Engineer | Tübingen, Germany | EDUCATION M.Sc. Machine Learning | University of Tübingen GPA: 1.9 (Current) — Tübingen, Germany Oct 2024 – Mar 2027 Focus: Generalist, GenAI, Synthetic Data B.Sc. Computer Science | University of Tübingen GPA: 1.8 / 1.0 — Tübingen, Germany Oct 2020 – Sep 2024 Final Thesis: Generating Professional Portraits from Amateur Photos via Personalization of Stable Diffusion Received Deutschlandstipendium PREVIOUS WORK Startup Engineer & Research Assistant (HiWi) | University of Tübingen — Tübingen, Germany Nov 2024 – Jul 2025 • Architected full-stack GenAI platform as the foundational engineer to productionize bachelor thesis. • Orchestrated k-shot inference on a HPC Slurm cluster (8×L40S, 16×RTX 3090) using Redis and WebSockets to manage 100k+ jobs with estimated wait time report, having a peak throughput of 6,000 generations per hour. • Developed a generative augmentation framework using InstantID and FLUX.1 Kontext, while improving facial resemblance via FaceNet, leading to a first-author publication at a CVPR 2025 Workshop. • Optimized generation latency to 14s by identifying bottlenecks via E2E profiling and implementing model pre- loading with a “warm-node” queue management system to eliminate cold-start latency. • Sustained a 49% activation rate (visit-to-generation) for 2,500+ visitors, supporting high-retention cohorts with average 40+ monthly generations per registered user and 14–22 minute sessions for power users. Stack: Python (FastAPI, PyTorch, wandb), SQLite3, Redis, Slurm, ComfyUI, Vue.js, Apache2, Linux (Self-hosted); GitHub Actions SELECTED PROJECTS Delayed Auditory Feedback Online | Real-Time DSP Web App (Micro-SaaS) Feb 2025 – Present • Identified an underserved niche in speech therapy, built a sub-6ms round-trip web audio engine featuring hardware-floor latency reporting, achieving latency parity with native applications and dedicated hardware. • Scaled platform to 700+ MAU and 1000h+ total monthly session time via 30% direct traffic, proving stickiness. • Directed SEO strategy, achieving a 2.4 weighted average position and 75% CTR on primary search intent. Stack: JavaScript (Web Audio API), HTML, CSS, Formspree; Google Search Console, Google Analytics Spurious Feature Annotation Platform | Distributed Dataset Auditing System Apr 2023 – Jul 2023 • Architected a modular HITL tool to identify spurious features with multi-source API dataset enrichment, enabling researchers to label datasets via a centralized backend to offload model inference. • Steered a 4-person Agile team as Technical Lead, managing the end-to-end SDLC via Scrum. Stack: Python (PyTorch, timm, torchvision, HuggingFace, gradio, NumPy), JavaScript, Conda; GitHub Actions/Projects PUBLICATIONS Ulusan, K., & Kiefer, B. (2025). Generating Synthetic Data via Augmentations for Improved Facial Resemblance in DreamBooth and InstantID. CVPR 2025 Workshop “Synthetic Data for Computer Vision”. COURSEWORK Lectures: Reinforcement Learning · Trustworthy ML · AI Safety · Benchmarks · Non-convex Optimization · NLP · LMs Gaussian Processes · Recurrent & Generative NNs · CUDA · C++ · DBMS (PostgreSQL/DuckDB) Selected Coursework Projects: • RL Paper: Implementing TD3 with PER / Reward Shaping / Truncated n-Step Return Using SB3 Feb 2025 • Paper: Understanding Disagreement in Peer Review Feb 2025 • Code: Training a VAE on 64×64 images to achieve latent space vector disentangling Dec 2024 • Presentation: Synthetic Image Generation with a focus on GANs Dec 2023 LANGUAGES: Turkish (Native) · English (Professional, IELTS: 7.0) · German (Advanced, DSD: B2/C1) HOBBIES: Homelab · 3D Printing · FreeCAD · Adobe Photoshop/Illustrator · Stained Glass Art · Karate (Brown Belt) [Link: https://korayulusan.github.io/] [Link: https://korayulusan.github.io/] [Link: https://github.com/KorayUlusan] [Link: https://github.com/KorayUlusan] [Link: https://linkedin.com/in/korayulusan] [Link: https://linkedin.com/in/korayulusan] [Link: https://scholar.google.com.tr/citations?user=DZqc5tkAAAAJ&hl=en] [Link: https://synthetic-face-augmentation.github.io/] [Link: https://korayulusan.github.io/delayed-auditory-feedback-online/] [Link: https://korayulusan.github.io/blog/i-built-delayed-auditory-feedback-online-tool/] [Link: https://arxiv.org/abs/2505.03557] [Link: https://arxiv.org/abs/2505.03557] [Link: https://github.com/KorayUlusan/Reinforcement-Learning-TD3/blob/main/hockey-technical-report.pdf] [Link: https://github.com/KorayUlusan/Understanding-Disagreement-in-Peer-Review/blob/main/datalit_project_report.pdf] [Link: https://github.com/KorayUlusan/Beginner-Guide-to-GANs/blob/main/Koray-Ulusan_Image-Generation.pdf]