RAG & Vector Search
Knowledge retrieval systems with thoughtful chunking, hybrid search, and grounded generation.
QdrantOpenSearchContext Engineering
AVAILABLE FOR ML CONSULTANCY & ARCHITECTURE
Independent engineer · Production-minded AI
Sergey
Kazanbaev
Senior AI/ML Engineer
I design and build practical AI systems that move beyond demos — from NLP and RAG to LLM orchestration and computer vision.
01 / What I do
Core expertise
Turning research-grade ideas into dependable systems teams can operate, improve, and trust.
LLM & Agents
Composable model workflows, efficient serving, and reliable agent architectures built for real tasks.
vLLMLangChain / LangGraphDistillation · RoPE
MLOps & Self-Hosted Infra
From model packaging and deployment to observable, resilient high-load inference services.
PyTorchDocker · KubernetesLinux · High-load serving
Computer Vision & Speech
Multimodal pipelines that extract useful signals from images, audio, and unstructured data.
Audio ASREntity ResolutionImage Pipelines
02 / Tools of the trade
Tech stack
A focused toolkit for building, shipping, and maintaining intelligent products.
PythonPyTorchHugging FaceCatBoostFastAPIDockerQdrantPostgreSQLLinux
03 / Selected builds
Featured projects
01 / RETRIEVAL
RAG Document Engines
Document ingestion and search pipelines designed for precise retrieval, source awareness, and useful answers.
Python· Qdrant · FastAPI
02 / KNOWLEDGE
Local AI Knowledge Bases
Private, self-hosted assistants that turn personal and team knowledge into a searchable workspace.
LLMs· Embeddings · Docker
03 / MULTIMODAL
Audio QA Pipelines
Speech transcription and analysis workflows that transform recorded conversations into structured insights.
ASR· NLP · Python
04 / Say hi to the shell
Interactive terminal
A tiny CLI for the curious. Try help.
sergey@ai-engineer: ~
zshWelcome to Sergey’s portfolio shell. Type help to see available commands.
Ready when you are.
QUICK RUN:
Have an interesting problem?
Let’s make AI useful.
Need a technical sounding board, an architecture review, or a partner to take an ML idea to production? I’d be glad to hear what you’re working on.
Start a conversation