> Case Loopt | Torno Studio

[← Back to projects](https://tornostudio.com/en/#quem)

Operational AI · Professional experience

# Loopt: from an ML model to yard operations

At Loopt, I worked on the core Machine Learning solution and later took on product management for the AI product. The challenge was to turn manual container allocation into a product used in daily operations.

## What does the Loopt experience teach about putting a model into operation?

A model needs to support the decisions people make while doing their work. At Loopt, I built the core model and later managed the container allocation product, connecting modeling, priorities and yard operations. This case describes professional experience predating the studio.

* My role

  ML engineering and product management

* The project

  AI product for container terminals

* The evidence

  Company results with a public source

[My contribution](https://tornostudio.com/en/cases/loopt/#participacao) [Results and evidence](https://tornostudio.com/en/cases/loopt/#resultados) [Sources](https://tornostudio.com/en/cases/loopt/#fontes)

● ● ●loopt.com.brActual screenshot

[![Screenshot of the public Loopt website describing sequencing, positioning and execution.](https://tornostudio.com/_astro/loopt-public-site.BxpzY5cx_107Pzo.webp)](https://tornostudio.com/_astro/loopt-public-site.BxpzY5cx.webp)

Loopt’s public website, captured October 14, 2026. This provides product context; it is not an internal screen or a claim that I built the website.

## A positioning decision shapes the work that follows

The product I worked on optimizes container allocation in the yard. A poor position can require extra moves when another unit needs to leave. The problem combines frequent decisions, operational constraints and scattered information.

My role brought together model development and the work of turning its output into a useful operational decision. Delivery needed to extend beyond modeling into everyday use.

My contribution

## From ML engineering to product leadership

01

### Building the core

As a Machine Learning engineer, I developed the core solution from modeling to production using Machine Learning and Deep Learning.

02

### Product management

I then took on the PM role, connecting operational needs, priorities and development with a cross-functional team.

03

### Staying hands-on

I continued writing production code. This experience connects product decisions with their implementation and operational consequences.

## From data to an operational decision

1. 01

   ### Operational data

   Information from systems already used at the terminal.

2. 02

   ### Optimization

   The product evaluates conditions to recommend yard decisions.

3. 03

   ### Operational use

   Recommendations support positioning and execution.

A conceptual view based on the public product description, not a reproduction of internal architecture.

The outcome

## Results reported by Loopt

According to Loopt in MundoLogística on January 13, 2026, the technology reduced unproductive moves by 27% and increased productivity by 21%.

Published result · according to Loopt

Normalized indices: each indicator starts at 100.

### Unproductive moves · −27%

* Baseline

  100

* With the solution

  73

03570105140

### Productivity · +21%

* Baseline

  100

* With the solution

  121

03570105140

100 → 73 represents a 27% reduction; 100 → 121 represents a 21% increase. These are not absolute counts or a time series.

Results of the solution developed by the team, according to the company. They do not measure my individual contribution in isolation.

## What I bring from this experience

An AI product needs to connect an operational problem, modeling and execution. My experience at Loopt covers those stages: first building the ML solution, then leading its evolution as a product.

## Sources and context

My role is described from my professional history. This is experience predating the studio, not a Torno client engagement.

* [Loopt website ↗](https://loopt.com.br/)
* [MundoLogística · January 13, 2026 ↗](https://mundologistica.com.br/noticias/loopt-ia-terminais-de-conteineres)

Case study prepared October 14, 2026.

## Does your process need software that works in daily operations?

Let’s understand the problem, the available data and how to verify a first delivery.

[Book a free 45-minute diagnostic](https://calendly.com/henriquepett/45min)

Other projects

[Bom.Capital →](https://tornostudio.com/en/cases/bom-capital/)[Ascende →](https://tornostudio.com/en/cases/ascende/)