Relink - Transforming the way people are matched with jobs

Transforming the way people are matched with jobs

Relink

CONTACT [email protected]

PRESS INQUIRIES [email protected]

INVESTOR INQUIRIES [email protected]

ADDRESS Njalsgade 19D, 2300 København S, Denmark

• Just not efficient • Limitations to our processing capabilities • The talent ecosystem is changing fast - becoming hard to keep

up and understand context

Why transform? Some of the challenges

“it all starts and stops with people”

We decided to build a recommendation engine were the job descriptions becomes the search query.





• Profiles and job descriptions - augmentation • Interaction between “profiles and jobs” • Our solution: build network effect and create transparency -

recommendations as a service (or - a quick win in the “sexy category)

Our “super matchmaker” needs data, data and more data





A job is published in an ATS - our tech provides: • A feed of recommended candidates • Analysis of incoming applications

Ok, so what do we sell





Our recommendation capabilities and the data we harvest can (potentially) reinvent the whole experience of finding your next job or colleague

We believe that has huge implications

To make this happen - we are a data-first company

Data-first company





• Computer science background • PhD in Machine Learning • Worked in digital advertising tech, cyber-security & academia • Lead data scientist at Relink

• DS/ML architecture • Bridging the gap between DS and production • Product development

Bio - Bjarne Ø. Fruergaard





Product development

Idea

Measure

ImplementProfit

Machine learning





Product development

Idea

Measure

ImplementProfit

Machine learning

Idea

Measure

ImplementProfit





1. Hire a data scientist

A day in the life

3. Get disappointed

2. “Please improve our product, but do not interfere with delivery”





accessible, easy and fast should not interfere with delivery

modular pipelines model repository

execute (& train?) models live measure KPIs live A/B testing live

pending all of the above

• Gather data • Bulk read, long-running, resource intensive

• Apply modeling • To a lesser extend, develop models and/or algorithms

• Execute experiments • Often on a fixed dataset • Measuring for model selection, not KPIs directly

• Follow state-of-the-art (or invent it)

The work of a <insert-title-here>Data scientist, ML engineer, DS/ML/AI researcher, etc.





ArchitectureOr: How you learn to stop

worrying and love the data Part I

λ (lambda) architecture • Immutable event store • Functions over all data at once

Batch layer: Periodically updates functions over all data

Speed layer: Real time updates functions over latest data

Serving layer: batch + speed





λ• Immutable event store • Functions over all data at once

Batch layer:Periodically updates functions over all data

Speed layer:Real time updates functions over latest data

Serving layer:batch + speed

Operational benefits • avoid incremental state • fault tolerance • easier to scale horizontally

Data scientists • bulk and streaming access • no interference with delivery • create, extend and compose “functions”





CompetencesOr: How you learn to stop

worrying and love the data Part II

Efficient data science

Engineers: implementation, modularity, monitoring, tech stack, scalability, testability, etc.

Designers, social sciences, product owners, UX, UI

• throughput • stay up to date • ideas





• Never as simple as first thought • Ship code and start measuring • Be able to compare alternatives • Improve and invent

• Chickens and eggs • Long-term investment

• Needs to be core to business value • Can you experiment and measure?

• Understand what data scientists are not

• Support process before scaling DS • Keep an eye on tech

No free lunch :-)

Idea

Measure

ImplementProfit





We’re hiring :-)http://relinklabs.breezy.hr





Questions?





Contact us: Ståle Fredlund Husby, CEO, [email protected]

Bjarne Ø. Fruergaard, Head of Data Science, [email protected]

mailto:[email protected]

mailto:[email protected]

Relink - Transforming the way people are matched with jobs

Data & Analytics