Berlin · Available for new work

Adjoint ConsultingMachine learning tools
built for the problem at hand.

Scientific and consumer products from diagnosis to deployment.

Contact us→See the process↓
Process · define → test → deploy → scale
scroll
Process

From an unclear problem to a working system.

Some projects begin with a well-defined target. Others require careful framing before a technical approach can be chosen.

Across six stages, we work with you to define the problem, map the data and workflow, test a focused build, deploy it, and expand it when needed.

Stage 01 / 06

Problem Definition

Costs may be rising, decisions slowing, or a promising product idea may have uncertain feasibility. Clarify the problem, the known constraints, and the result needed before choosing an approach.
Stage 02 / 06

Workflow Map

Map the workflow, inputs, constraints, and current baseline. Check which data can be trusted and what can be tested without creating avoidable risk.
Stage 03 / 06

Solution Design

Choose a focused first version and define how it will be tested: a model, data pipeline, analysis tool, interface, or combination of these.
Stage 04 / 06

Pilot

Test the system in a controlled setting. Compare it with the baseline, record failures, and require human review where errors matter.
Stage 05 / 06

Deployment

Deploy it as a standalone tool or connect it to existing systems. Add monitoring and documentation, and assign responsibility for maintenance.
Stage 06 / 06

Scale

Increase users, data volume, capacity, and security controls as needed. Address bottlenecks before expanding further.
Work

Product development, system evaluation, and technical diligence.

Our work includes scientific machine-learning research, consumer software development, and technical review of medtech hardware and chemistry-based products.

01

Problem and Data Review

Define the problem, map the relevant data and workflow, and determine what is worth building.

02

Machine Learning and Software Development

Design and build a testable system for a specific scientific or consumer-product need.

03

Model and System Evaluation

Test robustness, analyze failures, and assess whether a model or system is ready for its intended use.

04

Deployment and Scale

Deploy a working system, add monitoring and maintenance, and expand capacity or functionality when needed.

05

Technical Diligence

Review technical claims, code, evidence, and literature for product, investment, or procurement decisions.

Selected public work

Gradient guided furthest point sampling for robust training set selection

A published sampling method for selecting molecular machine-learning training sets. The paper and Python reference implementation are public.

Read the paper ↗View the code ↗
About

Morris Trestman

I'm a technical consultant, machine-learning researcher, and chemist based in Berlin.

Clients work with me and my team directly; a project may cover problem definition, technical review, implementation, or deployment.

01LinkedIn/in/morris-trestman02GitHub/mosstrestman
Contact

Tell us what you're working on.

Send a few lines about the problem, what exists today, and what you need to decide or build.

moss@adjoint.tech→
Based in Berlin