Yusuf Sarıgöz
CTO & Co-founder at Altai.dev
Ankara, Ankara, Türkiye Turkey
Summary
🤩
Rockstar🎓
Top SchoolYusuf Sarıgöz is a CTO and co-founder with nine years of experience specializing in multimodal AI, semantic search, generative AI and vision-language models. He leads product-facing engineering at Altai.dev to simplify enterprise LLM customization and is a former senior applied scientist and AI lead across startups including Qdrant and De Jure AI. A Google Developer Expert in Machine Learning, he combines research-grade model work with hands-on systems engineering—contributing LLaVA integration, memory optimizations, and gguf improvements to the widely used llama.cpp project. He has deep experience in speech and ASR systems, adding multilingual support and TPU-friendly RNNT implementations to TensorFlowASR. Based in Ankara, Yusuf bridges research, open-source impact and startup execution, often focusing on making powerful model customization practical and reproducible. Collected translation and Android-development training informs his interdisciplinary approach to ML productization and developer tooling.
9 years of coding experience
8 years of employment as a software developer
Nanodegree, Android Development, Nanodegree, Android Development at UdacityMaster of Arts, Translation and Interpreting, Master of Arts, Translation and Interpreting at Hacettepe University
Github Skills (16)
llama10
machine-learning10
gff10
c-language10
gm10
tensorflow10
cprogramming-language10
gf10
automatic-speech-recognition10
python10
gfs10
speech-recognition8
computer-vision8
speech-to-text8
conformer8
Programming languages (12)
Github contributions (5)
ggml-org/llama.cpp
Jun 2023 - Feb 2025
LLM inference in C/C++
Role in this project:
ML Engineer
ML EngineerContributions:53 reviews, 30 PRs, 122 pushes in 1 year 7 months
Contributions summary:Yusuf primarily contributed to the integration and support of LLaVA, a multimodal model, within the llama.cpp framework. They implemented the necessary code for LLaVA, including image encoding and inference, and addressed related issues. The user also worked on optimizing memory allocation and fixing bugs. Further contributions included supporting BakLLaVA conversion and enhancing the underlying gguf library.
ggmlllama
TensorSpeech/TensorFlowASR
Dec 2020 - Jan 2021
:zap: TensorFlowASR: Almost State-of-the-art Automatic Speech Recognition in Tensorflow 2. Supported languages that can use characters or subwords
Role in this project:ML Engineer
ML EngineerContributions:23 commits, 4 PRs, 28 comments in 26 days
Contributions summary:Yusuf primarily contributed to the development and enhancement of the TensorFlow-based ASR system. They implemented support for the Multilingual LibriSpeech dataset, added pure TensorFlow support for the RNNT loss function, and fixed issues related to the loss implementation. Furthermore, the user worked on integrating and supporting TPU training within the project, which demonstrates an understanding of distributed training frameworks.
ctcspeech-recognitionstate-of-the-artstreaming-transducertensorflow
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.