16
Palantir PALANTIR_DGTAXUD 19-02-2019 Ref. Ares(2020)3255548 - 23/06/2020

19 02 2019 - asktheeu.org

  • Upload
    others

  • View
    11

  • Download
    0

Embed Size (px)

Citation preview

Palantir

PALANTIR_DGTAXUD

19-02-2019

Ref. Ares(2020)3255548 - 23/06/2020

This material is confidential, intended for recipient

only, and should not be distributed.

Based in P

alo Alto,

with a global presence

Approx. 2500 em

ployees,m

ostly engineers

Started out in

government in 2004

Who are w

e? The facts.

This material is confidential, intended for recipient

only, and should not be distributed.

Who are w

e? The m

ission.

Laying out the “data foundation” for the m

ost important organizations.

Data circulation through an enterprise requires a

large number of transform

ations and integrations that need to be m

anaged transparently.

Palantir

PA

LA

NT

IR F

OU

ND

RY

This material is confidential, intended for recipient

only, and should not be distributed.

What do com

panies without a

data grid look like?

1) Data S

ilos “I can’t m

ake decisions based on the right data“

2) Ineffective interaction with data

“I have access to some data but I can’t ask the second, third, or

Nth questions to inform

operations”

3) Mistrust in D

ata “W

hose data is right?”

4) Inability to deploy strategy effectively“I can’t deploy and track strategies across m

y organization”

This material is confidential, intended for recipient

only, and should not be distributed.

Challenge 1

Data suffers from

diseconomies of scale –

the m

ore of it there is, the harder it is to make it

operational

Challenge 2

Data has to m

eet people where they are –

and am

plify their expertise

Foundry addresses two critical

organizational challenges.

Workflow

s

Ontology

Data Foundation

Foundry Core

This material is confidential, intended for recipient

only, and should not be distributed.13

DATA PR

EPARATIO

ND

EVELOPM

ENT

DEPLO

YMEN

TUS

AGE

MAN

AGEM

ENT

SEAR

CH &

ACCESS

Direct connection to the

customer’s S

3 infrastructure

and the CDL

Easily explore data sources for desired datasets

Schedule data extracts w

ith configurable alerts to m

onitor

data quality in each update

This material is confidential, intended for recipient

only, and should not be distributed.14 S

EARCH

& ACCES

SD

EVELOPM

ENT

DEPLO

YMEN

TUS

AGE

MAN

AGEM

ENT

DATA PR

EPARATIO

N

Quickly apply project-specific business

logic to drive feature development

Enable pipeline transparency and auditability through autom

atic data

provenance

Branch both data and code to enable

feature development and testing at

full scale

This material is confidential, intended for recipient

only, and should not be distributed.15 Interactive code notebook experience w

ith dynamically scalable resources

Easily configure the distributed com

pute ecosystem w

ith self-

service package managem

ent

SEAR

CH &

ACCESS

DATA PR

EPARATIO

ND

EPLOYM

ENT

USAG

EM

ANAG

EMEN

TD

EVELOPM

ENT

Templatize discrete m

odeling steps to curate a library of reusable,

trusted modules

This material is confidential, intended for recipient

only, and should not be distributed.16 Prom

ote the production model to a protected

branch and compare to historic versions

Build out-of-the box and and custom

visualizations to evaluate m

odel performance

Easily interpret stages of the deployed model

SEAR

CH &

ACCESS

DATA PR

EPARATIO

ND

EVELOPM

ENT

USAG

EM

ANAG

EMEN

TD

EPLOYM

ENT

This material is confidential, intended for recipient

only, and should not be distributed.17

Directly surface results of m

odel in operational applications

Interface with m

odel results in the ontology layer w

ith easily configurable visualization com

ponents

SEAR

CH &

ACCESS

DATA PR

EPARATIO

ND

EVELOPM

ENT

DEPLO

YMEN

TM

ANAG

EMEN

TUS

AGE

Evaluate model effectiveness in the field

by leveraging usage metrics &

integrated user feedback

This material is confidential, intended for recipient

only, and should not be distributed.18 M

anage complexity through

automated build schedules for

production pipelines

Understand m

odel inputs, logic, and dow

nstream usage

SEAR

CH &

ACCESS

DATA PR

EPARATIO

ND

EVELOPM

ENT

DEPLO

YMEN

TUS

AGE

MAN

AGEM

ENT

Manage m

odels as a step in the data pipeline w

ith model versioning, health

checks, and dependency tracking

Palantir

FO

UN

DR

Y IN

AC

TIO

N

This material is confidential, intended for recipient

only, and should not be distributed.29

This material is confidential, intended for recipient

only, and should not be distributed.30

-P

alantir is already driving leading-edge transform

ation across the custom

s and borders industry

-R

eal-time interventions on people

and goods, and breaking ground w

ith PIU

data (inc.AP

I/PN

R data)

-Transform

ing the way m

ajor shipping organisations operate, saving one organisation $100m

/year

Revenue collection

Transforming logistics

Targeting individuals and netw

orks

Facilitating trade

This material is confidential, intended for

recipient only, and should not be distributed.

Use

r G

ro

wth

at A

irb

us O

ve

r t

he

La

st 2

Ye

ars

Impact of data extends beyond

Airbus’ w

alls and into the larger industry

Since 2015, P

alantir has been adopted by over 50 airlines and reached 6,000 users across the A

irbus Skyw

iseplatform

, which allow

s operators to carry out predictive m

aintenance on planes and keep aircraft safely in the air for longer.

Operating at scale

31

Client S

ME

s, analytical/technical users, and IT engaged at the start; P

alantir team w

orking alongside client to accelerate delivery and

‘prove the platform’.

Palantir involvem

ent is decreasing as the platform

becomes established and internal pow

er-users, consum

ers, and IT gain experience and autonomy.