From the sensor to the ledger.
- 01 Cloud platforms and delivery
- 02 Data engineering
- 03 AI agents with real access
- 04 Control and connected systems
Otomatik is a small consulting practice working from the control system up to the cloud. Cloud platforms, data pipelines, AI agents, and the industrial systems underneath them, built by an engineer who has shipped on both sides of that boundary.
Four disciplines, one head.
Every line below is work already shipped in production. Anything outside this list, we will say so and point you at someone better.
Cloud platforms and delivery
Serverless AWS architectures, multi tenant on day one, provisioned entirely as code. Pipelines authenticate with short lived OIDC credentials, so a deployment never depends on a shared key in someone’s terminal.
Data engineering
Ingestion, modelling and governed sharing. On one platform this took client onboarding from three weeks down to 48 hours, with data refreshing every 15 minutes instead of overnight.
AI agents with real access
Context aware assistants wired to your own APIs, with an explicit limit on what they may change in a single turn. Useful enough to save real time, constrained enough to hand to staff.
Control and connected systems
PLC logic, HMI work, industrial protocols and edge provisioning. This is the half most software consultancies quietly subcontract, and the reason projects that cross the OT boundary tend to stall.
Small on purpose.
No account manager in front of an offshore team. You brief the engineer, and the engineer writes the code.
The two halves of industrial software are usually solved by two people who never meet.
Leon started at the hard end. CANBUS diagnostics and J1939 replay rigs, PLC logic in structured text to IEC 61131, HMIs for reclaimers and stackers on fixed plant, laser trackers and sensor fusion for robotics. When the code is wrong there, something physical stops moving. That sets a standard you carry into every project afterwards.
The second half came from data and cloud. Lead data engineer on a secure data sharing platform, then architecting the decomposition of a monolith into domain driven microservices for a national finance group, leading a distributed squad of six and putting behaviour driven testing in front of every release.
Notes from the work.
Written during live engagements. Specific, occasionally opinionated, never produced in bulk.
Multi tenancy is an IAM problem, not a database one
From three weeks to 48 hours, onboarding clients with data shares
Agents that do the work, and the one mutation rule
Tell us what is broken.
A short call, an honest read on whether we are the right fit, and a scope you can act on either way.
