Anton

Anton Podolian
Backend Developer | AI & Data Systems
Warsaw / Remote | Email: [открыть контакты](см. выше в блоке «контактная информация») | Phone: [открыть контакты](см. выше в блоке «контактная информация»)
Telegram: @mm3ttt4 | GitHub: https://github.com/serenitymea | LinkedIn: https://www.
[открыть контакты](см. выше в блоке «контактная информация»)

Summary
Backend Developer with 2+ years of experience building backend systems, ML pipelines, and AI-powered tools.
Focused on practical products at the intersection of media automation, data annotation, and generative AI.

Experience

Co-founder & Backend Developer — Splay Clipmaker Startup / Side Project
2025 – Present

• Founded and built Clipmaker (@splayvideoeditor_bot) — an automated beat-synced video editing pipeline deliv-
ered as a Telegram bot, reaching 100–150 active users per week
• Implemented audio beat analysis with librosa and video motion scoring with OpenCV to algorithmically select
and align the best fragments to music rhythm
• Built a user task queue with daily generation limits, file-based locks, and session state management
• Stack: Python 3.11, python-telegram-bot, librosa, OpenCV, ffmpeg, yt-dlp, Docker

Backend Developer | ML Pipeline NDA Project
2025 – 2026

• Built a human-in-the-loop data collection system based on pre-trained speaker diarization model outputs, coordi-
nating a pool of ∼90 freelance annotators
• Developed a backend API for serving model predictions, managing audio data, and storing user corrections
• Designed an interface for validating model outputs, where users listened to audio segments and corrected speaker
labels
• Implemented tools for adjusting segment boundaries and refining annotation quality
• Developed a task assignment mechanism for distributing audio samples among users for annotation
• Integrated asynchronous processing for handling model inference outputs and data preparation

Python Developer Freelance
2024 – Present

• Developed an automated TikTok edit generation tool that creates dynamic, trend-driven edits with minimal manual
input
• Built a Pinterest-to-SDXL image generation tool with Gradio UI for reference-based image synthesis
• Created a RAG QA system with FastAPI, VoyageAI, Gemini, and ChromaDB, enabling question answering over
custom text corpora with source references
• Built a real-time face swap pipeline with tracking and blending
• Developed a coaching services website with Django and PostgreSQL for course management and client interaction
Technical Skills

Languages: Python (3.10+), Java, JS, SQL
Frameworks: FastAPI, Django, PyTorch
Databases: PostgreSQL, SQLite
Testing: pytest, unittest
Media: ffmpeg, OpenCV, librosa, yt-dlp
Other: REST, JWT, asyncio, Linux, SQLAlchemy, Docker

Education
Warsaw University VIZJA — LLM and Data Science

Stack
Python, Backend, FastAPI, Django, PostgreSQL, REST API, Asyncio, SQLAlchemy, pytest, PyTorch, transformers,
diffusers, NumPy, OpenCV, FFmpeg, librosa, Telegram bots, Docker

Languages
English – B2
Polish – B1
Ukrainian – Native
Russian – Native