Sales Automation (PostgreSQL, PostgreSQL, Redis, Pydantic, Docker, curl_cffi)

Looking for a Python Middle/Middle+ developer for a one-time implementation of an event-driven platform, with a database-first approach and queues on Redis Streams.

Attached a .jpg file with an example of the data flow and processing (demo version)

Tasks:
- Configuration on Pydantic, structured logging.
- Secure storage of sensitive data: AES-256-GCM.
- Connections: PostgreSQL + Redis + MongoDB
- Network layer: working through proxy/fingerprints (client layer)
- Implement services: Session Manager, Session Agent/Puller, Data Normalization, Event Dispatcher, Task Scheduling, Telegram Gateway.
- Integrations: First integration (polling/HTML event source) - translate from aiohttp to curl_cffi. Second integration (GraphQL + WebSocket) - ready. Specifics of endpoints/platforms will be provided after selecting the contractor.

Functional requirements:
1. Management of integration sessions
2. Event reception (Agent Puller / Transport)
3. Event Bus on Redis Streams
4. Data normalization (using open source/my developments)
5. Event dispatcher (Routing)
6. Handlers/plugins
7. Task scheduler
8. Data storage
9. Configuration and management

Non-functional requirements:
1. Performance and scalability
2. Reliability of delivery (at-least-once)
3. Security (AES-256-GCM)
4. Resource efficiency (Predictable degradation)

Acceptance criteria:
- Repository with a modular structure
- Working launch via docker-compose (postgres/redis/mongo + services/workers).
- Flow: Raw event -> events:ingress -> Normalizer -> events:normalized -> Event dispatcher -> handlers
- Scheduler: placing delayed-task in ZSET and transforming it into an event for processing.
- Project documentation: how to set up, how to add a new handler/integration, event format (more details at the acceptance stage)

Stack (expecting experience):
- Python 3.11+; asyncio (preferably uvloop), background workers
- Redis Streams (consumer groups), Redis ZSET (scheduler)
- PostgreSQL (connection pools), MongoDB for logs/analytics
- Pydantic, Docker/Compose
- Experience with curl_cffi Async Session (retries/timeouts/proxy) is welcome

Work format:
- One-time task, with possible long-term cooperation
- Project details and test accounts after the start of cooperation
- Bi-weekly report on completed work. Picture/Video for media warming.

Response:
1. Briefly: how would you organize consumer groups, retries, errors, compromises of event order (global vs by integration_id) (1-2 paragraphs).
2. When you are ready to start work and an estimate for completion.

If the response is template-based, it will be skipped.