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

Block or report dkleptsov

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.

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
dkleptsov/README.md

Denis Kleptsov — AI/ML Engineer

Typing SVG


Website LinkedIn GitHub Profile views


👋 About

I build AI/ML systems end to end — from agentic LLM applications and RAG pipelines to applied ML and the full-stack apps that ship them. I work across the whole spectrum: cloud APIs when they're the right tool, and privacy-first, on-device models when the data should never leave the machine.

  • 🤖 Agentic LLM systems — multi-step agents and tool use with LangGraph, traced and evaluated in LangSmith; retrieval-augmented generation over private documents and the live web.
  • 🧠 Applied ML — PyTorch / Hugging Face, quantised local models (llama.cpp, Apple Silicon / Metal), speech synthesis & voice cloning, and ML-powered document/data pipelines.
  • 🎙️ Voice & conversational AI — text-to-speech / voice cloning and real-time audio/video experiences.
  • 🖥️ Full-stack delivery — FastAPI services and React / Svelte / Tauri front-ends, packaged with Docker; native macOS apps powered by on-device LLMs.
  • 🔒 Privacy by design — on-device inference and PII redaction when privacy matters: no cloud round-trips, no data leaving the machine.
  • 🌍 Languages: 🇬🇧 English · 🇷🇺 Russian · 🇪🇸 Spanish (learning).
  • 💬 Ask me about agents, RAG, evaluation, local LLMs, and shipping ML into real products.

🛠️ Tech Stack

Languages

Python TypeScript Rust Swift Shell

LLM & Agents

LangChain LangGraph LangSmith OpenAI Anthropic Ollama Chroma

Machine Learning

PyTorch Hugging Face ONNX llama.cpp NumPy

Backend & Infra

FastAPI PostgreSQL Redis Docker Linux

Frontend & Desktop

React Svelte Vite Tauri SwiftUI

Voice & Realtime

TorchAudio WebRTC WebSockets


🚀 What I build

Domain What Core tech
🤖 Agentic LLM apps Multi-step agents, tool use, RAG over docs + web, traced & evaluated LangGraph · LangChain · LangSmith · OpenAI / Anthropic · Chroma
🧠 Applied ML Quantised local models, speech synthesis / voice cloning, ML data pipelines PyTorch · Hugging Face · ONNX · llama.cpp (Metal)
🎙️ Voice & conversational Text-to-speech / voice cloning, real-time audio/video TorchAudio · WebRTC · WebSockets
📄 Document & data AI Translation, PII redaction, extraction & scraping with LLMs ScrapegraphAI · Playwright · PyPDF / docx / pptx
🖥️ On-device desktop apps Native macOS apps running fully offline Tauri · Rust · SwiftUI · local LLMs
🧩 Full-stack & infra Async APIs + reactive UIs, containerised FastAPI · React / Svelte · Docker · Postgres / Redis

Selected public work

  • 🧠 agentic-rag-assistant — an agentic RAG assistant on LangChain · LangGraph · LangSmith (FastAPI + React, streamed, traced & evaluated).
  • 🦫 Bobr Translate — an offline MS Office document translator for macOS, shipped on the Mac App Store.

📊 GitHub Stats

GitHub streak
Activity graph

Agentic AI · applied ML · privacy-first on-device — across cloud and Apple Silicon

Pinned Loading

  1. agentic-rag-assistant agentic-rag-assistant Public

    Agentic RAG assistant — LangChain · LangGraph · LangSmith (FastAPI + React, traced & evaluated)

    TypeScript 1

  2. auto-debug-agent auto-debug-agent Public

    An autonomous AI agent that writes Python code, catches its own errors, and self-corrects.

    Python