Daniil Makeev

AI Engineer

I build and ship AI agents and RAG systems end to end.

Open to full-time or contract · remote / relocation
About

I'm an AI engineer based in Moscow, open to remote or relocation across EU/US timezones. I started in Python backend in 2023 — ETL, data pipelines, internal services — and moved into applied AI in January 2025.

For the past year-plus I've been an AI engineer at Hao, shipping production RAG and content-generation pipelines, where I led a video pipeline to ×ばつ faster output at 60% lower cost. On the side I build and ship my own tools — an open-source browser agent, a multi-tenant contact RAG, a personalized news digest. Graduating Bauman MSTU (CS) in 2026.

Experience
2025 — Present

Led the video-generation pipeline end to end — dialogue generation, TTS, image compose, cost instrumentation — and shipped supporting RAG work and the marketing site. Rebuilt a linear orchestrator into a two-stage CLI with a JSON contract, parallel TTS + pluggable image generation, FFmpeg compose with karaoke subtitles, and per-generation cost tracking.

×ばつ faster output · −60% generation cost · batch mode for continuous runs

PythonLLM pipelinesRAGElevenLabs TTSFFmpegS3
2023 — 2025
Python Developer · Zenit-Elektro

Metrology company · internal tooling

Owned Python services across ETL and data quality for internal catalogs. Automated manual moderation across product catalogs and added validation across data imports from partner systems.

×ばつ faster site moderation · −60% import errors

PythonFastAPIAsyncIOPostgreSQLETL
View full résumé
Projects

A Chrome side-panel agent that runs multi-step tasks inside your real browser session — no headless cloud, no credential copy-paste. LangGraph.js plan-and-execute with per-tool state budgets, tab isolation, prompt-injection guards on every third-party text source.

Live on the Chrome Web Store · open-source, Apache-2.0

TypeScriptLangGraph.jsMV3 service workerpuppeteer-coreReact

An AI assistant for personal and organizational networks. Multi-tenant with per-tenant RLS isolation, semantic search over embeddings, news parsing, and a reranking loop that grounds answers in source records.

Live on Telegram — you can message it yourself

RAGpgvectorrerankingaiogramSupabase

Reads your channels — fetching the article behind link-only posts — and writes one summary tuned to your focus. A deterministic pipeline produces the daily digest; a separate stateful chat agent (LangGraph, checkpointed to Supabase) answers follow-ups.

Self-hosted · running daily

LangGraphSupabase checkpointerOpenRouterPTBDocker
Skills
AI / MLRAG · Vector search · Reranking · Grounded generation · Structured outputs · LLM deduplication
FrameworksFastAPI · LangGraph · PydanticAI · aiogram · python-telegram-bot · deepagents
DataPostgreSQL · Supabase · pgvector · Pydantic v2 · AsyncIO · httpx
PlatformsOpenRouter · OpenAI · ElevenLabs · Jina AI · Backblaze S3
DevOpsDocker · docker-compose · GitHub Actions · uv · Linux
LanguagesPython · SQL · TypeScript · English B2 · Russian native · Chinese B1

The projects here I built, shipped, and keep running myself. If that's the kind of engineer your team is missing, message me on Telegram (or email).

AltStyle によって変換されたページ (->オリジナル) /