AI Engineer: Applied NLP & Knowledge Graphs

About Happeo

• Happeo is a Series B startup revolutionizing how organizations collaborate and communicate
• through our unified social intranet platform. We combine collaboration tools, knowledge sharing,
• and internal communications into one seamless solution that helps teams connect and stay
• aligned. We pride ourselves on our dynamic, collaborative culture that emphasizes delivering
• high-quality solutions while fostering professional development. No bureaucracy, just smart
• people building things that matter.

About the Role

• You'll be working with our development team to build Happeo's proprietary technology for
• intranet information management. The platform helps organizations find gaps, duplication, and
• outdated content, and keep the knowledge their teams rely on accurate and trustworthy. We're
• launching into Open Beta, and the next frontier is knowledge that people, and the AI systems
• they use, can actually trust.
• This role builds toward Compass, Happeo's new knowledge verification layer. As AI systems like
• Claude, Gemini, and ChatGPT increasingly answer from an organization's own knowledge,
• Compass checks whether that knowledge actually holds up, surfacing where it's duplicated,
• stale, or self-contradictory, and proposing fixes. The job isn't search, it's detection: you'll build
• the knowledge graph and graph-RAG that understand what the knowledge says and whether it's
• correct, not just which documents mention what.

Our Stack

• React/React Native frontends, Python/ Node.js/Java backends, running on GCP (App Engine,
• Cloud Run, Kubernetes, Cloud SQL, Firestore, VertexAI).

What You'll Do

• You'll stand up the knowledge graph and graph-RAG behind Compass from scratch; this doesn't
• exist here yet, and building it is the job. It's novel work: extracting and structuring an
• organization's knowledge so issues in it can be detected. You'll ship from zero to production with
• minimal oversight and real autonomy to define the technical approach, make the architectural
• calls, and drive direction. This isn't a detailed-specs role: you'll form opinions about what to
• build, how, and why it matters, bring in knowledge the team doesn't have yet, and actively
• spread it.

Your Typical Day

• Build information extraction pipelines that turn messy documents into structured facts: entities and the relationships between them
• Build claim extraction and entity resolution: pull atomic, verifiable claims from documents, and decide when two extracted things are the same entity
• Detect where knowledge is duplicated, stale, or self-contradictory, and prove it works without crying wolf
• Stand up the knowledge graph the detection runs on, and the graph-RAG layer that supports it alongside conventional RAG
• Own the impact end to end: ship, measure, iterate, fail fast, learn faster
• Evangelize what you bring in: level up the wider AI team so the knowledge sticks past you

What We're Looking For

The core of this role is applied NLP: turning messy, unstructured knowledge into verifiable

structure. That's where most of the work, and most of the difficulty, lives

• Information extraction from unstructured text: entity and relationship extraction, plus ontology and taxonomy development, turning messy documents into structured facts
• Claim extraction and fact-checking: pulling atomic, verifiable claims out of documents (e.g. "SLA = 72h") and reasoning about whether they conflict (natural-language inference), are stale, or are unverified. This is the hard part of "self-contradictory" and a graph-RAG-for-search background often won't have touched it.
• Entity resolution and deduplication: deciding two extracted things are the same entity and canonicalizing them; spotting when two documents are versions of each other
• Strong, practical experience applying NLP and LLMs to real problems
• Knowledge graphs and graph databases, required as the foundation the detection runs on: Neo4j or similar, fluent in Cypher, and you've built a graph-RAG system
• Backend experience ( Node.js or Python preferred) and cloud, ideally GCP
• Comfortable building for multilingual content: our users and their knowledge aren't all in English
• A doer, not an academic: the work needs real research, but you ship and prove fast and make calls with incomplete information, rather than going deep and slow
• Upbeat, high-energy, extreme ownership: you own the impact, not just the task

Nice-to-have

• Docker/Kubernetes
• Java backend experience
• Data engineering and ETL pipelines
• DevOps/CICD experience (MLOps/LLMOps)

What Success Looks Like in 90 Days

This role has a clear early bar, and it's an ambitious one. By day 90, success looks like

• A knowledge graph stood up and live in production: the foundation Compass runs its detection on, not a prototype
• Working detection: Compass can flag where knowledge is duplicated, stale, or self-contradictory, with claims and entities resolved well enough to trust
• Measured on precision and recall, not ranking: you're confident you're surfacing real issues, and just as confident you're not raising false ones
• The wider AI team has learned something from how you built it: you've brought knowledge in and spread it

The home run is detection people trust: real issues caught, false positives kept low enough that teams act on what Compass tells them. The misfire is a clever graph that flags noise nobody believes. We're hiring for the first one.

Location

• This role is based in Helsinki, Finland . The working mode is Hybrid. If needed, we'll support you
• with relocation and Visa application.

Perks and Benefits

• A competitive salary and equity plan
• Flexible office setup with remote and hybrid working possibilities
• Full 5-weeks holiday policy from day one
• Comprehensive benefits, including healthcare, lunch, and phone subscription
• Top-tier work equipment you need to succeed
• A strong karaoke culture at practically every company event!

• _______________________________________________________
• Happeo is committed to unbiased recruitment, ensuring that all prospective employees,
• regardless of background, gender, race, age, or any other characteristic, are given equal
• consideration. We value different perspectives and encourage people of all identities,
• experiences, and abilities to apply. If you're passionate about the role and meet the
• qualifications, we'd love to hear from you; your unique background and experiences are
• welcome here.
• If you like us but this job isn't quite right for you, please submit an open application through our
• website. You may be the perfect candidate for our next opening!