Machine learning & data engineering

Machine learning systems that survive contact with production.

We design, build and ship data and machine learning systems — from the first prototype that proves the idea to the pipeline that keeps it running afterwards.

An ensemble of decision trees Three decision trees drawn as conifers. A single highlighted path runs from the root of the central tree down through one branch to a leaf, representing one prediction in the ensemble.

Services

What we take on

Most engagements start in one of these and grow into the next. We are happy to own a whole system or to slot into a team that already has one.

  • Machine learning engineering

    Model development, training pipelines, hyperparameter search, and evaluation you can defend in a review.

  • Data platforms & pipelines

    Ingestion, storage and transformation that stay correct when volume grows and schemas shift underneath you.

  • Web & application development

    Production web applications and the APIs behind them, built to be handed to someone else later.

  • Technical consultancy

    Architecture reviews, feasibility work and second opinions — usually before a budget gets committed.

  • Technical documentation

    Decision records, runbooks and handover material written for the engineer who arrives after you.

  • Project & delivery management

    Scope, milestones and visible progress, for teams without a technical lead to spare.

How we work

Four steps, in this order

The sequence matters more than the labels: we spend money on the risky part first, while changing direction is still cheap.

  1. 01

    Scope

    A call to understand the problem, then a written scope: what we will build, what we will not, and how we will both know it works.

  2. 02

    Prototype

    The riskiest component first. If the hard question has no good answer, you find out in week two rather than month five.

  3. 03

    Build

    Working software in front of you throughout, with direct access to the people writing it. No reveal at the end.

  4. 04

    Hand over

    Documentation, runbooks and a walkthrough, so your team can run and change the system without us in the room.

Why work with us

We specialise in taking on the awkward parts

  • Short feedback loops

    We ship early and often, without trading away the things that make a system maintainable a year later.

  • Direct access

    You talk to the engineers building your system. There is no account-management layer between you and the work.

  • Applied AI experience

    Reinforcement learning, distributed data collection and model tuning, across both research and commercial settings.

What we build

The shape of the work

Representative examples of the kinds of systems we design and ship, drawn from our own work and research. Client engagements are covered by NDA and are discussed on a call.

  • Distributed systems · ML

    Real-time scraping at scale

    A coordinated network of agents collecting data in real time, feeding a training pipeline that fine-tunes a model with hyperparameter search layered on top.

  • Reinforcement learning · Security

    Adversarial resilience testing

    Two competing reinforcement learning agents simulating a denial-of-service failure mode, to find where a system gives way before someone else does.

  • Reinforcement learning · Networks

    Routing & spectrum allocation

    Custom Gymnasium environments for routing and wavelength assignment in optical networks, with policies trained against realistic traffic.

  • Reinforcement learning · Simulation

    Control policies from simulation

    Agents trained in simulation and moved towards production-shaped environments, from arcade-game benchmarks through to control problems that matter.

Get in touch

Have a project? Tell us what you are trying to build.

Send a few lines about the problem — what exists today, what should exist instead, and roughly when. We will reply with questions, an honest view on feasibility, and a suggested next step.

Email us

or write directly to info@randomforest.solutions

Based in
London, United Kingdom
Working with
Clients worldwide, remote