Latest publications
All publications →- Efficient Semantic Retrieval via Multilingual Embeddings and RerankingYılmaz, R. E., Taysi, M. A., Özmen, A. İ. , İnce, G. — International Symposium on Innovations in Intelligent Systems and Applications (INISTA 2025) — to appear (2025), 2025Semantic RetrievalRerankingMultilingual
- Grounded Answer Generation over Multimodal Financial Records via Semantic IndexingYılmaz, R. E., Taysi, M. A., Özmen, A. İ. , İnce, G. — 10th International Conference on Computer Science and Engineering (UBMK 2025), IEEE, 2025
- Exploring vocal biomarkers as non-invasive fine-tuning assays of cardiovascular health: heart failure modelYılmaz, M. B., Durmus, M., Yılmaz, R. E., Polat, Z. P., Colakoglu, S. — European Heart Journal, 45(Supplement 1), ehae666–987, 2024
From the blog
All posts →- Oct 9, 2025How to Build High-Performance LLM Systems?
A practical, end-to-end guide to making LLMs feel instant: what to measure (TTFT, TPOT, P99), where the time goes (prefill vs. decode, KV cache), and which optimizations actually move the needle (PagedAttention, FlashAttention, quantization, batching, caching) — with a chatbot-style pipeline as the running example.
LLM InferencePerformanceLatency - Oct 9, 2025Mastering Multi-Vector Embeddings: Beyond Traditional Semantic Search
Multi-vector embeddings represent a paradigm shift in semantic retrieval — moving from single compressed vectors to fine-grained token-level understanding. This article explains the architecture, math, and practical optimizations behind modern multi-vector systems.
Information RetrievalEmbeddingsColBERT - Oct 9, 2025MUVERA: Making Multi-Vector Retrieval as Fast as Single-Vector Search
Google Research’s MUVERA bridges the gap between multi-vector and single-vector retrieval, offering near-ColBERT accuracy at a fraction of the latency. Here’s how it works and why it matters for the next generation of semantic search systems.
Information RetrievalEmbeddingsMulti-Vector
Let’s collaborate
Enterprise search, multilingual IR, and multimodal/RAG systems. I’m open to projects, research partnerships, and talks.