Skip to content
View tahsinsoyak's full-sized avatar
🎯
Focusing
🎯
Focusing

Block or report tahsinsoyak

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.

Content in all repositories owned by your account will be closed.
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
tahsinsoyak/README.md

Tahsin Soyak

🤖 AI/ML Engineer & Researcher · İstanbul, Turkey


LLM agents: tools, retrieval, and multi-agent systems LLM infrastructure: benchmarking, routing, and deployment Applied research: domain LLMs and computer vision


About Me

I am an AI/ML Engineer & Researcher based in İstanbul, turning large language models and computer vision research into high-availability production systems.

  • AI Engineer @ DT Cloud: Building autonomous LLM agents (planner / retriever / writer / evaluator), an end-to-end LLM Benchmark Platform for Türkiye's Defense Industries Agency (SSB), and self-hosted multi-agent (A2A) architectures.
  • MSc in Computer Engineering (GPA 3.96 / 4.00): Thesis on Fine-Tuning Large Language Models for Turkish Agriculture.
  • TÜBİTAK Research Project: Fine-tuned Google Gemma with LoRA/PEFT, successfully passing benchmark-driven milestone evaluations.
  • Published IEEE Researcher: Computer vision & applied ML · 6× Academic Honour Certificate recipient · Former Microsoft Student Ambassador.
  • Open to research collaborations and discussions on LLM agent swarms, evaluation harnesses, and domain fine-tuning.

Browse all repositories →  ·  Read my articles on Medium →

Tech Stack

  • AI & Models: Python · PyTorch · TensorFlow · LangChain · LiteLLM · Hugging Face · OpenCV
  • Backend & Systems: FastAPI · Django · Docker · Redis · Celery · PostgreSQL
  • Architectures & Methods: RAG · LoRA / PEFT Fine-Tuning · A2A Multi-Agent Swarms · Qdrant / Chroma · Benchmarking (EM/F1)

Experience

  • AI Engineer · DT Cloud Jun 2025 - Present · İstanbul
    Architected the LLM Benchmark Platform for SSB (Türkiye's Defense Industries Agency) using Django, LiteLLM, and Redis/Celery. Built self-hosted A2A multi-agent platforms with latency/cost-aware routing. Fine-tuned Gemma (LoRA/PEFT) for agricultural intelligence under a TÜBİTAK grant.
  • R&D Engineer · Mirsis Information Technologies Sep 2023 - Jun 2025 · İstanbul
    Engineered transformer-based CV parsing for an intelligent recruitment platform, migrating legacy NLP into fine-tuned LLM services in Django.
  • Founder & Captain · Familiar Nov 2022 - Present
    Built meeting summarizers, content generation assistants, and medical-imaging classification pipelines (breast cancer and brain tumour).
  • Scientific Researcher · University of Groningen Aug 2023 - Oct 2023 · Netherlands
    Engineered custom machine learning models and visualizations for scientific research publications.

Research & Publications

  • IEEE (2025): Strawberry Counting & Ripeness Detection on the Web (Real-time web computer vision pipeline)
  • MSc Thesis (GPA 3.96 / 4.00): Fine-Tuning Large Language Models for Turkish Agriculture (TÜBİTAK funded)
  • Applied ML (2023): Song Popularity Classification Using the Spotify Dataset
Verified Certifications (Click to expand)
  • Agents & Retrieval: Hugging Face (MCP · AI Agents) · NVIDIA (RAG Agents) · LangChain (Chat with Your Data)
  • Models & NLP: Google (Gemini Multimodal) · DeepLearning.AI (Fine-tuning LLMs · NLP in TensorFlow)
  • ML & Edge: Stanford (Machine Learning Specialization) · IBM (ML with Python) · NVIDIA (Jetson Nano)

📊 GitHub Analytics

GitHub Contribution Streak

🏅 Leadership & 🌍 Languages

  • 🧠 Founder & President: AI Society, Yozgat Bozok University · 🎤 Microsoft Student Ambassador · 🏆 6× Academic Honour Certificate recipient
  • 🌍 Languages: 🇹🇷 Turkish Native · 🇬🇧 English Fluent · 🇳🇱 Dutch Elementary · 🇩🇪 German Elementary

Contact

Looking to collaborate on agent systems, fine-tuning, or production LLM infrastructure?

Email LinkedIn

Thanks for visiting! Feel free to star the repos or connect.

Pinned Loading

  1. song-popularity-prediction song-popularity-prediction Public

    Positioned the project as a foundation for future research in song feature analysis, with implications for personalized music recommendations and enhanced user experiences on music streaming platf…

    Jupyter Notebook

  2. breast-cancer-detection breast-cancer-detection Public

    This project implements a comprehensive breast cancer classification system using neural networks with multiple evaluation methodologies. The system analyzes breast cancer data to classify tumors a…

    Python

  3. strawberry-ai-on-web strawberry-ai-on-web Public

    A Flask-based web application that analyzes images of strawberries using YOLO object detection models to identify and classify strawberries by their ripeness state.

    HTML

  4. rag-pdf-chat-app-hf rag-pdf-chat-app-hf Public

    A Retrieval Augmented Generation (RAG) application that allows users to chat with their PDF documents. Upload a PDF and ask questions about its content to get accurate, contextual answers.

    Python