Helsinki, Uusimaa, Finland · Hybrid
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 commun…
Skills: Applied NLP, Knowledge Graphs, Graph-RAG, Information Extraction, Entity Resolution
Sign up with Clera and we'll reach out the moment a role actually fits you — no more spraying applications into the void.
Build a knowledge graph and graph-RAG system from scratch to detect duplicated, stale, or contradictory information within organizational knowledge. Develop information extraction pipelines to turn unstructured documents into structured facts and verifiable claims.
Requires strong practical experience in applied NLP, LLMs, and knowledge graphs using tools like Neo4j and Cypher. Candidates should have backend experience in Python or Node.js and be comfortable working with GCP and multilingual content.
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outdated content, and keep the knowledge their teams rely on accurate and trustworthy.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!
Curious about Happeo? Join one of our weekly drop-in product tours: bit.ly/dropintour
Happeo is a next-gen intranet that helps teams manage knowledge and internal communications in one unified place. Offering a template-based page builder, as well as integrations and universal search across all company tools, Happeo is easy to use and scale for companies of any size. That's why market leaders like Doctolib, GANT, and Marqeta rely on Happeo to keep their teams informed, aligned, and productive. With adoption rates 3x industry average, Happsters are on a mission to help organizations grow in extraordinary ways. See how they're doing at happeo.com.
Offices: Pursimiehenkatu 26 C, Helsinki, Helsinki 00150, FI · Singel 373, Amsterdam, North Holland 1012WL, NL · 500 7th Avenue, New York, New York 10018, US · Constantin Brancusi 25, Cluj-Napoca, 400458, RO
Curious about Happeo? Join one of our weekly drop-in product tours: bit.ly/dropintour
Happeo is a next-gen intranet that helps teams manage knowledge and internal communications in one unified place. Offering a template-based page builder, as well as integrations and universal search across all company tools, Happeo is easy to use and scale for companies of any size. That's why market leaders like Doctolib, GANT, and Marqeta rely on Happeo to keep their teams informed, aligned, and productive. With adoption rates 3x industry average, Happsters are on a mission to help organizations grow in extraordinary ways. See how they're doing at happeo.com.
Offices: Pursimiehenkatu 26 C, Helsinki, Helsinki 00150, FI · Singel 373, Amsterdam, North Holland 1012WL, NL · 500 7th Avenue, New York, New York 10018, US · Constantin Brancusi 25, Cluj-Napoca, 400458, RO
