Giorgio Sidari

ML & Software Engineer

Custom software

I help businesses develop their products and processes, supporting them in make-or-buy evaluations, selecting vendors, designing and realizing proprietary software.

Process analysis

Solution design

Product discovery

Team management

Full-stack development

R&D

Let's build your business advantage

I help you beat your competition by distilling scientific research into ad-hoc, distinctive technologies powering your products.

A few past projects:

Portable engineering

Mobile app to support the sales of small photovoltaic projects. 3D CAD, power and optimal setup calculation. So the engineering team can focus on big plants.

Advanced routing

Navigation engine for pickup and delivery of multiple asset categories with different resupply policies.

Knowledge extraction

Document indexing and linking to provide efficient access to the organisation's data lake.

Conversational technologies

Chatbots and assistants improving communication within your organization and with your customers. Based on your organization's needs, we can leverage either or both:

Cloud services

Natively scalable solutions with guaranteed availability, provided by leading companies like Google, Microsoft, Amazon.

Open-source software

Cutting-edge technologies allowing for full data ownership and greater accuracy.

Mastodon

AI policy

AI-powered workflows can dramatically accelerate software development, but they come with choices about where your code and data are processed. I give my clients full control over this decision.

On-premise workflow

All processing happens exclusively on the client's own systems: local machines or private servers under their direct control. No code, data, or project context is transferred to third-party infrastructure. This is the right choice when intellectual property, personal data, or regulatory constraints demand it.

Outsourced workflow

Processing leverages cloud-based AI services from authorised providers, with temporary transfer of project material to remote servers. This unlocks the most capable AI models, enabling faster delivery and broader technological reach, ideal when speed and capability outweigh data residency concerns.

The choice is made together during project setup and can be revisited at any time. Different parts of the same project can follow different workflows: speed where it's safe, control where it counts.

For a deeper look at these workflows and the questions to consider, see Speed and Safety in AI-assisted Software development.