Dawid Bartczak / Working together
Process automation and system integrations
I connect tools and automate repetitive data workflows. Explore my approach, examples from BILO and InVault, and experiments with receipt OCR.
Let’s discuss your projectWhen it makes sense
If people copy the same data between tools, check statuses or run tasks on a schedule, it helps to describe each step. Automation is useful when the expected action, source of data and person responsible for exceptions are clear.
Possible scope
Data flow
Retrieving information, validating its format and passing it to another system. Feasibility depends on the available APIs, exports and permissions.
Tasks and events
Actions triggered by a status change or a schedule. Retries, preventing duplicate execution and visible error history are important parts of the design.
Documents and input data
OCR and information extraction begin with a sample of documents. Uncertain results may need human review; a model should not approve data without checks.
Where we start
We begin with sample data and the source and destination systems. We check technical access, workflow rules and failure cases before choosing the integration method. Verification uses agreed examples, including interrupted connections and receiving the same event twice.
Examples of my work
BILO connects sales, payments and ticket handling. InVault runs backups on a schedule and when a drive is connected. Paragon Pipeline is a separate experimental OCR/NLP project: it demonstrates receipt data extraction research, not a production system or verified accuracy on a client’s documents.
Before we begin
Can every application be integrated?
No. APIs, export options, vendor restrictions and data access need checking. Without that information, the scope and maintenance of the connection cannot be estimated reliably.
Does automation require AI?
No. Clear rules can be implemented with ordinary code. A model may help with unstructured content, but its quality and uncertain outputs need separate evaluation.
Describe your project
Name the tools you move data between and give one example task. How often do you do it, and what should happen if the data is incorrect?
Email Dawid