Data platform
Industrial data foundation
Pipelines that carry the plants’ operation — production, weighers, quality — into a single, reliable data warehouse.
Software and data for the feed industry, from the plant floor to the decision.
Over four years, I designed and built, in partnership with AGPR5, the data platform and the Gemba product suite that serves dozens of animal-nutrition plants in Brazil. It brings the plants’ operation — dosing, weighing, quality — into one place, measures the precision of each recipe and delivers decisions through dashboards, reports and AI assistants.
Role · Architecture, data engineering, data science, full-stack development and applied AI
Scope · Product suite · ~4 years · in production
In short, for decision-makers
The problem
Animal-nutrition plants generate data across many fronts — production, dosing, quality, commercial — scattered across plant-floor and management systems that don’t talk to each other, hard to turn into decisions.
What LNData did
Over four years we built a data platform and the Gemba product suite, covering several fronts: from dosing precision to digital quality (BPF), from production indicators to commercial intelligence and AI — delivered as multi-tenant products to dozens of plants through AGPR5.
The result
Each front became a trackable number and a product that scales: less dosing waste, standardized quality, commercial decisions backed by data and AI assistants — the whole operation in one place, reliable and up to date.
2022–present
four years of partnership
Gemba
proprietary product suite
Animal nutrition
feed industry
dozens
of industrial units
10+
products in the Gemba suite
3
generations of the data platform
end-to-end
from source to dashboard
multi-tenant
data + AI SaaS
Architecture
The platform integrates each client’s plant-floor and management sources, consolidates everything into an analytical data warehouse and delivers the result through several channels: BI, proprietary products and APIs.
Client sources
- Gemba MES
- SCADA
- Gemba BPF 3.0 API
- ERP and CRM
- Gemba Core
Data platform
Pipeline orchestration
Analytical data warehouse
Containerized · multi-tenant
Consumption
- BI · dashboards
- Gemba Analytics 2.0 · proprietary portal
- Analytics API · read-only
Areas covered
A product suite that spans from raw plant-floor data to commercial intelligence — all anchored in the same animal-nutrition domain.
Gemba Stats
Dosing precision, measured
A plant-floor data warehouse that computes, batch by batch, how far each dose fell outside tolerance.
Gemba DRAI
Optimized dosing
Data science that recommends the best weigher parameters per silo and product — with statistical proof of the gain.
Gemba ASAP
Plant analytics
An assisted-dosing application and production indicators, installed inside each unit.
Gemba BPF 3.0
Digital good practices
Digital quality checklists (GMP/Kamishibai), integrated with the ERP and turned into indicators.
CRM & commercial
Commercial intelligence
A proprietary CRM and dashboards that cross sales, market and client records to prioritize opportunities.
Applied intelligence
Data science and AI in production — not as a proof of concept, but measuring dosing, optimizing the weigher and answering questions about the business.
Data science in dosing
Models that optimize ingredient weighing and prove, statistically, the reduction in deviation and waste.
AI that answers about the business
Conversational assistants that query the data respecting each user’s exact permissions — defense by role, not by prompt.
AI-powered support (RAG)
A proprietary help desk that answers from the platform’s knowledge base, with semantic search.
Product · Gemba Analytics 2.0
The data portal each client uses as their own
The evolution of BI: instead of a generic tool exposed to the client, a proprietary multi-tenant portal where each plant sees only its own data, adjusts parameters that feed back into its own calculations, generates PDF reports and talks to an AI assistant.
Multi-tenant with isolated data
Each client sees only its own operation, with data separation enforced at the database level.
The client’s own parameters
The client versions rules and targets that feed back into the calculation — the dashboard reflects their reality.
Reports and AI
PDF reports generated in the background and a conversational assistant to ask about the numbers.
Access governance
Access control by company and by role, applied across the whole data layer.
FlaskCeleryReactClickHouse
Impact
The operation of dozens of plants, once trapped in isolated systems, now lives in a single, comparable data platform.
Dosing precision became a trackable number — the basis for reducing ingredient waste and standardizing feed quality.
Delivery evolved from artisanal analyses to multi-tenant SaaS products that scale to new clients with no rework.
AI and data science in production: dosing optimization with statistical proof and assistants that answer about the business safely.
Platform stack
PythonAirflowClickHouseReactDocker
A data challenge in industry?
From the source on the plant floor to a data product in the client’s hands — tell us what you need to solve and we’ll come back with scope, timeline and deliverables.