Machine Learning · Zürich

Applied AI for complex technical problems

Wedefin Labs builds production systems for teams with problems generic tools can't solve, applying AI where it earns its place, from satellite imagery to portfolio risk.

Email usSee our work
  • Swiss GmbH, Zürich
  • ETH-trained
  • Research-driven
  • Hands-on delivery

Team experience includes

ETH ZürichUC BerkeleyChalmersInnosuisseDeutsche BankHewlett Packard EnterpriseTelia CompanyBCGKaiju
ETH ZürichUC BerkeleyChalmersInnosuisseDeutsche BankHewlett Packard EnterpriseTelia CompanyBCGKaiju
ETH ZürichUC BerkeleyChalmersInnosuisseDeutsche BankHewlett Packard EnterpriseTelia CompanyBCGKaiju

How we work

Research depth, delivered as working systems.

Method

We start where the technical uncertainty is highest. Feasibility, data quality, model behaviour. We resolve it before anything gets built at scale.

That order matters. Most failed AI projects were architecturally sound and aimed at a problem the data could never support.

Zürich, seen across the Limmat

Based in

Zürich, Switzerland

Access

You work directly with the two people who build the system. No account layer, no handoff to a delivery team.

Scope

Fractional expertise, fixed-scope builds, or a focused advisory sprint when the question is which direction to take.

Services

How Wedefin Labs can help

Geospatial AI & Earth observation

Satellite imagery, remote sensing, and spatial analytics turned into decision-ready signal. Crop and land monitoring for agriculture, plus infrastructure, energy, and climate applications, built by a team that has shipped Earth observation systems in production.See Terrabit

Quantitative & financial risk systems

Portfolio analytics, protocol risk, and performance verification for teams that need rigour a generic dashboard cannot provide.See Wedefin

Agentic and research-grade AI systems

Autonomous agents, simulation, and decision support built by people who understand the underlying mathematics and science, not just the API.See our work

Sectors

  • Agriculture & Land Use
  • Space & Earth Observation
  • Climate & Energy Technology
  • Fintech & DeFi Infrastructure
  • Quantitative Finance
  • AI & Data Platforms
  • Scientific R&D

Selected case studies

Real products. Built from research to production.

Our own products are where we test the same capabilities we bring to client work: agentic AI, geospatial intelligence, quantitative systems, onchain infrastructure, and technical product execution.

Onchain product · Fintech infrastructure

Live product

Wedefin

Turning investment strategy into open, onchain competition.

A non-custodial crypto index protocol where builders create strategies, compete on verifiable performance, and help shape an index that anyone can access.

Technical product strategyEVM smart-contract systemsPortfolio and risk analyticsWallet and protocol integrations
Read case study Build something similar

Agentic AI · Physical-world intelligence

Live product

Terrabit Earth

An AI-native intelligence network for the physical world.

Autonomous analyst agents monitor research, public data, and Earth observation, then publish concise intelligence with sources, confidence, and verification attached.

Autonomous analyst agentsEvidence and claim verificationEarth-observation pipelinesGeospatial change detection
Read case study Build something similar

Autonomous analyst · Private deployment

Private deployment

Wedefin Sentio

A natural-language analyst over your own data, and the open web.

A retrieval-augmented analyst that answers questions in plain language across structured and unstructured sources, and runs web research when the answer is not already in house.

Agent harnesses and tool useRetrieval-augmented generationNatural-language query interfacesWeb research and source gathering
Read case study Build something similar

Quant · Automation · Compliance

Client work

Everything else

The work that never became a product of ours.

Forecasting and price models, quantitative research for trading desks, and automation across sales, compliance, and legal.

Discuss your use case

Team

Two people. Both build.

Wedefin Labs is deliberately small. The people who scope your problem are the ones who solve it.

Miller Mendoza, Co-Founder & CEO of Wedefin Labs

Miller Mendoza

Co-Founder & CEO

PhD Physics, ETH Zürich

  • Former CTO of a geospatial AI company, building Earth observation systems on satellite imagery and machine learning
  • Deep experience applying remote sensing to agriculture and land monitoring
  • Quantitative research and data science across risk and portfolio analytics
  • Co-founder of Wedefin, a live onchain investment platform
LinkedIn ↗
Sören Lambrecht, CTO of Wedefin Labs

Sören Lambrecht

CTO

MSc Data Science, ETH Zürich · Exchange and joint research, UC Berkeley

  • Enterprise generative AI delivery: agentic workflows and decision systems
  • Forecast modelling and price optimization for large industrial clients
  • Takes research-stage models through to production engineering
  • Previously P&L responsibility for a 1B+ SEK business unit at Telia
LinkedIn ↗

Engagement models

Flexible ways to work together

Advisory Sprint

A short, time-boxed engagement, typically one to two weeks, to answer a single question: whether an architecture, a product direction, or a technical approach is worth pursuing. Ends with a clear recommendation, not a maybe.

Fixed-Scope Project

A defined deliverable with the price and timeline agreed before work starts. A prototype, an analytical platform, a dashboard, or a model, built end to end and handed over.

Fractional Expert

Embedded, ongoing access to senior expertise for as long as you need it, part-time or full-time, without the overhead of a full-time hire.

What could we build together?

Start with a short intro call. We will look at the opportunity, name the hardest assumptions, and outline a path to production.

Email us

info@wedefin.com · Usually a reply within 24 hours