Let's talk.

Manufacturing software on Django and MySQL, automation of exported SAP reports in Python, Excel data pipelines, and AI agents on LLM APIs. Typical builds: role-gated factory systems for production and quality reporting, report suites that replace manual export-and-clean-up work, and agents that read production data and flag anomalies. If it removes manual work from a factory, it is in scope.

It is a fair question, and you should not accept a system only one person can read. So the stack is deliberately boring: Django, MySQL, Python. Widely used, heavily documented, and maintainable by any competent developer without specialist training. Your operational data sits in a standard relational database on infrastructure you control, in a documented schema. The value is in that data and it does not depend on me continuing to be available.

I am an industrial engineer who builds software, and I am specific about the difference. I am not a career senior developer. I know manufacturing processes, I know what the shop floor actually needs, and I build and ship working tools using AI assistance. More than 12 standalone tools are running in production. If you need deep computer-science engineering, hire a senior engineer. If you need someone who understands the process and can deliver working software against it, that is what I do.

AI is the implementation layer, not the judgment. Deciding what a plant needs, what the data model has to be, and what must never be allowed to happen is industrial engineering, and that part is mine. What AI changes is how fast a specified thing gets built and tested. That is why delivery is measured in weeks rather than quarters, and why one person can deliver what usually needs a vendor team. I state the method openly because you are entitled to know how the work is done.

Yes, and nothing inside SAP is touched. I automate the work that happens after the export: you drag in the raw report your team already pulls, and the tool parses it, stores it in a database, reconciles it, and produces the finished report in the layout managers expect, with date-range history. There is no integration into SAP, no transaction driving, and no change to your SAP system. One suite took daily reporting from roughly 2 hours of manual clean-up to under 4 minutes, with no copy-paste errors.

No, and deliberately not. I work from the reports your team already exports. There is no connection, no credentials issued to me, no add-on installed, and nothing changed inside SAP. That constraint is the feature: it removes the integration project, the security review and the risk to a system you cannot afford to have broken. One suite took daily reporting from roughly 2 hours of manual clean-up to under 4 minutes this way.

It reads production and downtime data, compares it against the plan, flags anomalies, and writes a short plain-language summary of the top issues with likely causes, delivered before the shift meeting instead of a shift too late. I build agents grounded strictly in real operational data, with no invented numbers.

Because I sign confidentiality agreements and keep them. Every project on this site is described generically for that reason, including the ones I would most like to show you. The same protection would apply to you. If a portfolio names its clients and quotes their internal numbers, that tells you what it would do with yours.

Yes. I am open to freelance, volunteer, and trial projects, and I am happy to start small so you can judge the work before committing to anything. The contact form is the fastest way to reach me. Tell me the manual process you want gone and your timeline.

Yes. Half my experience is inheriting systems and making them boring and reliable, security hardening, audit trails, documentation, then careful extension without stopping what already works.

It depends on the process, which is why the first step is a conversation rather than a price list. The shape is small, defined pieces of work rather than a multi-year platform programme, priced per build so you can stop at any point. I am open to freelance, volunteer and trial projects, which is the fairest way for both of us to find out whether this works before either commits to something larger.

That is the main focus. Smaller plants usually carry the most manual reporting because enterprise systems were priced or scoped out of reach, which is exactly where a focused tool pays for itself fastest. Engagements are scoped to one painful process at a time rather than a multi-year platform programme.

Usually yes. Most of what I build sits alongside systems that already work, reading their exports or extending them, rather than replacing them. I will tell you honestly if a codebase is beyond what I should take on, and I would rather say that up front than discover it halfway through.

With the process, not the software. You describe the work that is eating hours, I ask what the data looks like today and who touches it, and then I tell you whether software is the right fix. Sometimes it is not, and saying so costs you nothing. If it is, a small paid or trial build proves it before anything larger.