Andrey
Lapochkin

Python backend developer, ~7 years of experience, 50+ delivered projects. I build backends that keep working after launch. Recent work: - Multi-tenant VoIP/CRM SaaS: FastAPI, PostgreSQL, Redis pub/sub, WebSocket, LiveKit calls, arq workers, Terraform provisioning, push via FCM/HMS/APNs - Security audit and rescue of a legacy Django system: fixed data leaks, exposed DB, secrets in git - AI email agent for B2B outreach with LLM-generated replies - Telegram bots (aiogram 3) and MTProto clients, CRM integrations - Scraping and lead-gen pipelines with Playwright How I work: clear spec and staged estimates upfront, clean async code, Docker deployment, security by default. You talk directly to the person writing your code. Stack: Python, FastAPI, asyncio, PostgreSQL, Redis, Tortoise ORM, aiogram, Playwright, Docker, Terraform, GitHub Actions, LLM APIs

Work experience

Not specified

Education

Not specified

Andrey Lapochkin

(0)

I will review, debug and fix your Python backend (FastAPI, Django)

Web, Mobile & Software Dev

Python

$25

$25/hr

More info

MegaNET: lead generation & email outreach pipeline

End-to-end lead generation and email outreach pipeline controlled from Telegram. Parser bot: takes a search query, pulls results through SearXNG and SerpApi, filters out aggregators and low-quality domains, collects public contact data (emails, phones) and exports clean Excel reports via pandas. Fast path on httpx, fallback to Playwright for JavaScript-heavy pages. Smart Emailer: scheduled, personalized email sequences via Resend API that respect recipients' time zones and weekends, with load distribution to keep deliverability high. Access restricted to whitelisted users. Stack: Python 3.14, FastAPI, aiogram 3, Playwright, httpx, pandas, BeautifulSoup4, Docker Compose, Loguru.

DellCall: multi-tenant VoIP/CRM SaaS

Multi-tenant VoIP/CRM SaaS for distributed call centers with full branch isolation. Backend built from scratch: FastAPI, Tortoise ORM, PostgreSQL, Redis pub/sub, WebSocket notifications, arq background workers. Real-time calls via LiveKit, with a dedicated server per branch provisioned automatically by Terraform. Auth: JWT RS256 with refresh rotation and Redis-based token revocation. Push notifications across FCM, HMS, APNs and email through a shared provider-adapter layer. Mobile app on Flutter with VoIP calls via CallKit/PushKit; iOS releases through GitHub Actions + fastlane. My role: backend architecture, role-based access model, user-to-user calls, branch admin panel, tenant isolation hardening.

Zoom Funnel: webinar-to-meeting Telegram funnel

Telegram bot that converts webinar viewers into private Zoom meetings with zero manual work. An AI assistant guides users through the funnel, syncs available slots and sends reminders. Full integration with Zoom API and LPTracker CRM for end-to-end traffic tracking. Stack: aiogram, FastAPI, PostgreSQL (Tortoise ORM), Redis, APScheduler. Dockerized, with a CI/CD pipeline on GitHub Actions.

Big Shop Manager Bot: retail reporting automation

Telegram bot that automates reporting and discipline control across retail stores. Managers submit reports as text or voice (transcribed with Yandex SpeechKit), and everything lands in Google Sheets automatically. The bot reminds staff about deadlines, sends the owner a list of who missed their reports, handles out-of-hours auto-replies and tracks how fast managers respond to customers. The owner gets a daily summary for the whole team. Built on aiogram with FSM for branching scenarios, Google Sheets API, Docker. Result: no more manual chat checks, full transparency across all stores.

Smokey VPN: distributed VPN service backend

Backend for a VPN service: a Telegram bot for users and subscriptions, a central Master Node with the API and core logic, and agents on each VPN server collecting metrics. Supports single-server, split-service and fully distributed multi-server deployments. Monitoring with Prometheus and Grafana, automatic backups and rollback, YooKassa payments. Stack: Python, PostgreSQL (separate DB per service), Redis, Docker / Docker Compose / Kubernetes.

Solana Dev Sniper: new token alerts bot

Telegram bot that detects new meme tokens on Solana in real time via Helius API, filters and analyzes them (contracts, activity, basic metrics) and notifies users within moments of creation. Configurable alerts and filters. Architecture is split into handlers and workers to handle high load, with data stored for further analytics and trading automation. Stack: Python, FastAPI (webhooks), aiogram, PostgreSQL, Tortoise ORM, Helius API. Private project.

Shop Bot: digital goods store in Telegram

Telegram store for digital goods with fully automatic sales. Catalog with inline buttons, YooKassa payments confirmed in real time, instant delivery after payment. A separate FastAPI backend manages products and users without restarting the bot. Stack: Python, aiogram, FastAPI, YooKassa API.

RSMan: channel subscription gate bot

Telegram bot that gives access to content or features only to users subscribed to a given channel. Automatic subscription checks via Telegram API, support for multiple channels, easy to plug into any Telegram funnel. Used for private communities, paid content and subscriber growth. Stack: Python, aiogram, async processing.

AI Text-to-Speech Telegram bot

Telegram bot that turns text into natural-sounding speech with neural voices and sends back an audio file right in the chat. Async request handling for low latency, easy to extend with more voices and languages. Stack: Python, aiogram, neural TTS model.

Hereuknow CMS: custom portfolio engine

Custom CMS powering my portfolio site hereuknow.ru, with no off-the-shelf engines. Full CRUD for projects with categories and tags, rich-text case descriptions, built-in AI content generation with selectable LLM models, and view analytics. Backend: FastAPI (Python 3.14, uv), Tortoise ORM, JWT auth, Nginx, files on S3. Frontend: React + Vite.

AI Video Generator: Sora / Veo / Genmo

Service for generating AI videos from a text prompt through a single interface across several models (Sora / Veo / Genmo). The project started with Genmo, which had no official API, so I built the integration on top of its web client with browser-like request emulation. Other providers plug in through a common adapter, so new models can be added without touching the core. Generation jobs go through an async queue; the user gets a finished video when it's ready. Available as a Telegram bot. Stack: Python, FastAPI (async backend), httpx, async job queue.

AndreyLapochkin