I maintain systems.I connect technologies.I build automation.

Problem

Client contact, incoming requests and communication were scattered. A single place was needed to manage them.

What works

• →Public IT services website — mimax.com.pl
• →Client request form
• →Ticket numbering in the MX-xxxxxx format
• →CRM / helpdesk layer for handling cases
• →Text-based AI chat — available in the existing code

In progress / planned

• ○Voice AI — planned
• ○Further Telegram and Make integrations

Problem

Original writing needed a faithful, calm presentation and a simple publishing process without altering the author’s text.

What works

• →WordPress website structure and UX
• →Private presentation of original writing
• →Publishing workflow that preserves the author’s text

In progress / planned

• ○Planned audio content (ElevenLabs)
• ○Publishing automation

Problem

Job listings, multiple CV versions, application statuses and schedules were spread across different tools, making context and organization difficult.

What is built

• →FastAPI + PostgreSQL backend in Docker Compose, with a management panel
• →Application workflow and job listing import
• →CV versions linked to applications

In progress / planned

• ○Calendar and email automation
• ○Integrations — at varying levels of readiness
• ○Every application requires a human decision — no bulk, autonomous CV submission

Problem

Room measurement and installation planning need accurate spatial data; manual measurements are slow and error-prone.

What is built

• →Room scanning with LiDAR and Apple RoomPlan
• →2D and 3D views of scanned spaces
• →Designer module and Reality / texture pipeline

In progress / planned

• ○App stability
• ○Depth and measurement accuracy
• ○Scan coverage
• ○Designer → Reality synchronization

Problem

Tickets, client cases and reminders needed a single view that tracks deadlines rather than relying on memory.

What is built

• →Ticket and client records
• →Telegram alerts and notifications

In progress / planned

• ○Additional modules and automation added in stages
• ○Some views still use demo data — kept separate from working functionality

Problem

Testing which combinations of automation and local AI models provide real value before putting anything into use.

What is built

• →Make / n8n workflows controlled via Telegram
• →Local AI and ComfyUI on a dedicated GPU workstation

In progress / planned

• ○Iterative testing and pipeline development — prototypes

Problem

In-conversation support is useful only when it is fast, local and processes short time windows — without sending everything to the cloud.

What is built

• →Polish / English transcription prototypes
• →Analysis of CVs and job descriptions
• →Processing in short time windows

In progress / planned

• ○Low latency and local operation
• ○Consolidating both tools — planned, not a finished product