Founding AI Engineer

Overview

The construction sector experiences massive annual losses amounting to $1.6 trillion, primarily due to inefficiencies that could be avoided. Although the knowledge to prevent these losses exists, it remains locked away in inaccessible formats like PDFs, meeting notes, and email threads, causing each new project to relearn lessons already discovered. Our mission is to create software tailored for managing complex, large-scale construction projects by effectively interpreting documents, tracking key decisions, and identifying errors early on. Unlike competitors, our strength lies in a growing project memory that improves continually with each project completed.

Currently, our system operates on critical infrastructure projects including an autobahn highway and an S-Bahn transit program, handling multi-year schedules and hundreds of thousands of technical documents and communications. We are at the pre-seed stage with backing from prominent investors and over 25 active clients.

Key Responsibilities

• Developing agent harness engineering for construction document processing, catering to varied interpretations such as "structural risk" across different document types like RFIs, cost reviews, and schedule reconciliations.
• Building multi-agent pipelines that route extensive documents and archives into specialized evaluation streams with accurate contextual understanding.
• Designing systems for context compression to manage information retention and relevance over multi-year projects involving numerous stakeholders.
• Constructing and evolving a decision graph capturing project decisions including rationale, decision-makers, alternative options, and outcomes to benefit future projects.
• Utilizing a modern tech stack including TypeScript, Next.js, Vercel platform, Supabase database, LangChain, and AI SDKs with opportunities to employ advanced AI tools and models.

Work Culture

• Focus on shipping products with 80% readiness and refining them iteratively based on live project data.
• Preference for reading research papers over relying on existing libraries, as most solutions are cutting-edge and novel.
• Open and constructive feedback culture aimed at improving outcomes rapidly.
• High responsibility levels where individuals own problems end to end and deliver solutions.

Candidate Profile

• Ability to translate user challenges into actionable technical solutions with measurable impacts.
• Comfortable engaging with non-technical users to understand their workflows.
• Competent in front-end technologies, specifically TypeScript and React, alongside AI engineering expertise.
• Strong prototyping skills with an emphasis on measurement and reliability in production environments where errors have significant consequences.
• Self-driven in navigating uncharted problems by conducting research, building prototypes, and iterating solutions without established best practices.
• Diverse experience levels welcome, from fresh graduates to senior engineers, with scope tailored to individual capability.
• Preferred but nonessential qualifications include substantial open-source contributions, previous startup early-team experience, deep knowledge in document understanding or agent systems, or projects that significantly reduced human labor.
• Working language is English; German knowledge is beneficial but not mandatory due to native-speaking team members.

Compensation and Benefits

• Critical role as a founding engineer owning major product components and collaborating closely with company founders.
• Equity offer ranging from 0.5% to 1.5%, vesting over four years with a 1.5-year cliff.
• Hybrid work model requiring three days per week in Munich's central office, negotiable in special cases.

Hiring Process

• First stage: 30-minute interview with an engineering team member.
• Second stage: 30-minute discussion with the technical co-founder.
• Final stage: Onsite case study in Munich involving preparation and a collaborative session to evaluate problem-solving and coding skills.
• Final decision communicated within 24 to 48 hours post-case study.