Peak of Data & AI Encore: Scalable Design & Infrastructure

When your workflows need to scale across teams, systems, or infrastructure, thoughtful design is everything. In this final encore webinar, we bring together top sessions from the Peak of Data & AI that address how to build for the future: structured schemas, efficient architectures, and scalable systems: all powered by FME.

Talk 1: Unlocking FME Flow’s Potential: Architecture Design for Modern Enterprises

From small-scale setups to enterprise-level deployments, this session walks through key architectural decisions that impact performance, cost, and scalability. Learn how to build flexible infrastructure that grows with your organization.

Talk 2: Using FME and Excel to Develop Schemas and Generate GIS Feature Classes and Tables

See how Excel can bridge communication between GIS and non-GIS teams by visualizing and validating schema designs. With a few extra columns, the same Excel file becomes a blueprint for FME to generate feature classes and tables – speeding up collaboration and delivery.

Talk 3: Pixel Puzzler: Extracting Value From Images with Computer Vision and GenAI Techniques

Discover how to turn your growing archive of image data into structured, actionable insights. This session explores how FME integrates with computer vision and pre-trained ML models to automate tagging, connect unstructured imagery to CRM and DAM platforms, and tackle “dark data” at scale. Plus, you’ll see the “power of the prompt” in action through a fun Generative AI demo, all hosted on FME Flow.

Join us to sharpen your design skills and equip your organization for long-term success with FME.

1. Unlocking FME Flow’s
Potential:

Architecture Design for 

Modern Enterprises
The Peak of Data

and AI 2025

2. 2025
The Peak of Data and AI
Jean-Nicholas Lanoue

GIS Strategic Advisor
Since its inception in 1989, our GIS firm has been driven by a singular mission: 

to help organizations unlock the full potential of their geospatial data
Théo Drogo

GIS Strategic Advisor

3. 2025
The Peak of Data and AI
Unlock the full potential of
FME Flow by making the
right architectural decisions,

whether you're deploying it
for the first time or scaling it
across the enterprise.

4. 2025
The Peak of Data and AI
In Scope (We will cover…)
▪Architecture Fundamentals
▪Express, Distributed or Fault-Tolerant?
▪Focus on Engines
▪Key architecture decision points
▪Real-world architecture examples
Out of Scope (20 minutes only !)
▪FME Flow Hosted
▪Infrastructure As Code
▪FME Flow Administration concepts
▪Containerization
▪Licensing models
▪And everything you won’t see in the next slides "

5. 2025
The Peak of Data and AI
Architecture Fundamentals
▪5 essential components
▪Each play a specific role in execution & orchestration
▪Applies to all deployments
… Only 5, but they scale in many ways
Web Application
Server
FME Flow Core FME Engine FME Flow
Database
FME Flow
System Share

6. 2025
The Peak of Data and AI
▪Express
3 main deployment options:
Express, Distributed, Fault-Tolerant

7. 2025
The Peak of Data and AI
3 main deployment options:
Express, Distributed, Fault-Tolerant
▪Distributed

8. 2025
The Peak of Data and AI
▪Fault-Tolerant
3 main deployment options:
Express, Distributed, Fault-Tolerant

9. 2025
The Peak of Data and AI
Engine deployment options
▪Same server, same network
▪Different server, same network
▪Different server, different network
Engine type in 1 word!
▪Standard engine : classic
▪Remote Engine: nomad
▪CPU-based engine: scalable
Focus on Engines

10. 2025
The Peak of Data and AI
Cloud
On-prem
Database
Linux
Windows
Hardware
ArcGIS
Web Server
Compatibility Performance
Other important considerations

11. 2025
The Peak of Data and AI

12. 2025
The Peak of Data and AI
Key Architecture decision points
BUSINESS TECHNOLOGY
Time to 

value
Criticality
Scalability
Compliance & Governance
Budget
Business 

cases
Internal 

expertise
Data

location
Security
Infrastructure
Availability
Technology

stack
Interoperability

13. 2025
The Peak of Data and AI
Real-life scenario - #1
Industry context
Local gov (population = 150k). GIS is a mature department.
BUSINESS TECHNOLOGY
Business cases
5–10 FME users

Low volume of geo and tabular data

Not mission-critical
Infrastructure
Fully on-premise environment

All apps deployed on internal VMs
Budget
Restricted budget 

Cost-efficiency is a priority
Technology stack
Windows-only backend

ESRI 100% (Portal, Online, SDE, etc.)

SQL Server
Scalability
No short-term scaling needs

FME is not the corporate ETL
Data location
Internal file servers and database

Public APIs
Compliance &
Governance
Standard local gov compliance rulesAvailability No high availability or failover required
Internal expertise
5 GIS technicians, supported by IT 

New to FME but tech-comfortable
Security
Deployed behind corporate firewall

SSL enforced for all services

14. 2025
The Peak of Data and AI
   Distributed Architecture
oDatabase on SQL Server
oCPU Engine (1) on core server
oStandard Engine (1) on existing ArcGIS Server – for ArcGIS data integration
Real-life scenario - #1
Industry context
Local gov (population = 150k). GIS is a mature department.
BUSINESS TECHNOLOGY
Business cases
5–10 FME users

Low volume of geo and tabular data

Not mission-critical
Infrastructure
Fully on-premise environment

All apps deployed on internal VMs
Budget
Restricted budget 

Cost-efficiency is a priority
Restricted access to ESRI
licences
Technology stack
Windows-only backend

ESRI (Portal, Online, SDE)

SQL Server
Scalability
A few scaling needs for non-
spatial use

FME is not the corporate ETL
Data location
Internal file servers and database

Public APIs
Compliance &
Governance
Standard local gov compliance
rules
Availability
No high availability or failover
required
Internal expertise
5 GIS technicians, supported by IT 

New to FME but tech-comfortable
Security
Deployed behind corporate firewall

SSL enforced for all services

15. 2025
The Peak of Data and AI
Real-life scenario - #1
Industry context
Local gov (population = 150k). GIS is a mature department.
BUSINESS TECHNOLOGY
Business cases
5–10 FME users

Low volume of geo and tabular data

Not mission-critical
Infrastructure
Fully on-premise environment

All apps deployed on internal VMs
Budget
Restricted budget 

Cost-efficiency is a priority
Restricted access to ESRI licences
Technology stack
Windows-only backend

ESRI (Portal, Online, SDE)

SQL Server
Scalability
No short-term scaling needs

FME is not the corporate ETL
Data location
Internal file servers and database

Public APIs
Compliance &
Governance
Standard logal gov compliance
rules
Availability No high availability or failover required
Internal expertise
5 GIS technicians, supported by IT 

New to FME but tech-comfortable
Security
Deployed behind corporate
firewall

SSL enforced for all services
  IT add-ons
oOn prem VMs on Windows for the core, web server and CPU engine
▪6 vCPU + 32 Go RAM + 2 drives (OS vs App)
oTrusted SSL and SSO enabled
oDedicated service account

16. 2025
The Peak of Data and AI
Real-life scenario #2
Industry context
Large gas company. FME is widely used across the organization..
BUSINESS TECHNOLOGY
Business cases
Large volume of data

Real time field collection

Mission critical
Infrastructure
ESRI Enterprise On-prem

Asset management on AWS

Ongoing cloud transition
Budget
FME subscription 

Centralized FME budget (IT & GIS)
Technology stack
Hybrid OS

ArcGIS 100%

PostgreSQL
Scalability
Continuous growth on asset

Plan to expand to other geolocation
Data location Mix (cloud, on-site, SaaS)
Compliance &
Governance
Strong corporate governance rules

Industry compliance regulation
Availability
99.99% 

( < 1 hour down time per year)
Internal expertise
Strong GIS admin team

10 years of FME experience

IT & GIS work together
Security
Security team involved

ISO27001 obligation

17. 2025
The Peak of Data and AI
Industry context
Large gas company. FME is widely used across the organization..
BUSINESS TECHNOLOGY
Business cases
Large volume of data

Real time field collection

Mission critical
Infrastructure
ESRI Enterprise On-prem

Asset management on AWS

Ongoing cloud transition
Budget
FME subscription 

Centralized FME budget (IT & GIS)
Technology stack
Hybrid OS

ArcGIS 100%

PostgreSQL
Scalability
Continuous growth on asset

Plan to expand to other geolocation
Data location Mix (cloud, on-site, SaaS)
Compliance &
Governance
Strong corporate governance rules

Industry compliance regulation
Availability
99.99% 

( < 1 hour down time per year)
Internal expertise
Strong GIS admin team

10 years of FME experience

IT & GIS work together
Security
Security team involved

ISO27001 obligation
Real-life scenario - #2
Fault-tolerant Architecture
oDatabase on PosgreSQL on AWS
oRemote engines (2) on dedicated ArcGIS server only for licensing
oStandard Engines (4) for non-GIS data manipulation

18. 2025
The Peak of Data and AI
Industry context
Large gas company. FME is widely used across the organization..
BUSINESS TECHNOLOGY
Business cases
Large volume of data

Real time field collection

Mission critical
Infrastructure
ESRI Enterprise On-prem

Asset management on AWS

Ongoing cloud transition
Budget
FME subscription 

Centralized FME budget (IT & GIS)
Technology stack
Hybrid OS

PostgreSQL
Scalability
Continuous growth on asset

Plan to expand to other geolocation
Data location Mix (cloud, on-site, SaaS)
Compliance &
Governance
Strong corporate governance
rules

Industry compliance regulation
Availability
99.99% 

( < 1 hour down time per year)
Internal expertise
10 years of FME experience

IT & GIS work together
Security
Security team involved

ISO27001 obligation
Real-life scenario - #2
IT add-ons
oAWS on Linux for the core, web server and standard engines
▪8 vCPU + 32 Go RAM + 2 drives (OS vs App)
oOn-site VMs on Windows for the remote engines
▪4 vCPU + 32 Go RAM + 2 drives (OS vs App)
oSSL offloading of certificates through the loadbalancer

19. 2025
The Peak of Data and AI
“Why use one engine when you can
deploy five… in three networks…
and regret it later.”
— ChatGPT 4o on a Tuesday PM

20. 2025
The Peak of Data and AI
Thank You!
Curious to learn more? 

Be sure to attend our next presentation at 3:00 PM!
Navigating FME Migrations: 

The Essentials Before, During, and After
Ian Gagnon-Renaud
GIS Consultant
Laurent Phaneuf

GIS Consultant

21. Developing
Schemas with
FME & Excel
The Peak of Data
and AI 2025

22. 2025
The Peak of Data and AI
Mark
McCart
ETL Developer/FME Flow Admin
SAIC

23. 1.Introduction
2.Solution Inspiration
3.Tools Used
4.Wrap-Up
Game Plan

24. Solution
Inspiration
Section 2

25. Explaining database schemas and
tracking field changes can be
challenging when the audience
doesn’t have a GIS or database
background.

26. 2025
The Peak of Data and AI
Excel is Familiar
●Easy to track schema mapping when
migrating data from one schema to another
●During warehousing tasks, a single Excel File
can track field name/type changes for multiple
tables
●Easy to re-order attributes, add notes and
other important annotation during the
migration process that can be used for final
documentation

27. 2025
The Peak of Data and AI
Typical Setup

28. 2025
The Peak of Data and AI
Typical Setup

29. Tools Used
Section 3

30. 2025
The Peak of Data and AI
Esri and FME Tools
●Esri: Generate Schema Report Tool
●REST Endpoints -JSON
●FME: SchemaScanner Transformer

31. 2025
The Peak of Data and AI
I
=CONCAT(IFS(H2="String","fme_varchar("&I2&")",H2="Integer","fme_int32",H2="Double","fme_decimal(38,8)",H2="Date",
"fme_datetime",H2="Long","fme_int32",H2="Long Integer","fme_int32",H2="BigInt","fme_int64",H2="Short
Integer","fme_int16",H2="SmallInt","fme_int16",H2="GUID","fme_buffer"))

32. 2025
The Peak of Data and AI

33. 2025
The Peak of Data and AI

34. 2025
The Peak of Data and AI

35. 2025
The Peak of Data and AI
Attribute Aliases
●Import ‘Attribute Values’

36. 2025
The Peak of Data and AI
Attribute Aliases
●Import ‘Attribute Values’

37. 2025
The Peak of Data and AI
Attribute Aliases
●Import ‘Attribute Values’

38. 2025
The Peak of Data and AI
Writer Configuration
●FGDB doesn’t need exist
●Set Coordinate System
●Use ‘Dynamic’ FC/Table Definition

39. 2025
The Peak of Data and AI
Writer Configuration
●FGDB doesn’t need exist
●Set Coordinate System
●Use ‘Dynamic’ FC/Table Definition
●Set Feature Class Name to
‘fme_feature_type_name’

40. 2025
The Peak of Data and AI
Next Steps
●Editor Tracking
●GlobalIDs
●Domains/Subtypes
●Attribute Rules

41. Wrap-up
Section 4

42. 2025
The Peak of Data and AI
Wrap-up:
●Developing and/or migrating schemas require detailed
documentation
●Spreadsheets are a useful tool for this type of task
●Have fun with FME, try something new with the
platform to see if it can solve one of your data problems

43. 2025
The Peak of Data and AI
Thank You
Mark McCart

44. Pixel Puzzler:
Extracting value
from images
Avineon Tensing
The Peak of Data
and AI 2025

45. 2025
The Peak of Data and AI
David
Eagle
Director of Service Delivery
Avineon Tensing

46. In 1981, Bill Gates said that…
“640kb of RAM ought to be enough
for anyone”
He was wrong!
https://en.wikipedia.org/wiki/Bill_Gates

47. In 1999, Dale Lutz said that…
“FME will NEVER do raster”
He was REALLY wrong!

48. 2025
The Peak of Data and AI
Number of supported data types in FME
1993 2000 2007 2010 2015 2017 2020 2025…
10
100
300
500+
GIS
CAD
Database
XML
Raster
3D
BIM
Web
Point Cloud
Cloud
Big
Data
IOT
Gaming
BI
Indoor
Mapping
AR/VR
Generative
AI
Cloud
Native
Tabular
The passage of time…

49. 2025
The Peak of Data and AI
You might be expecting…
Translate Transform Resample Reproject Mosaic
Compress Clip Tile Drape Calculate Shade Appearance

50. 2025
The Peak of Data and AI
But, actually…

51. Chat logs
Email archives
Video footage
Raw survey data
IoT Sensor data e.g. temp
Social media posts
Call centrelogs
Site\survey photos
DARK DATA

52. Untapped Potential
Outdated or inaccurate
Costly –storage & environmental
Unnecessary backup durations
Legal and compliance issues
Hard to access
Awkward to search
Frustrating!
DARK DATA

53. DARK DATA
***PHONE STORAGE IS FULL***
Me: How can it be full already?
My camera roll:

54. DARK DATA

55. 2025
The Peak of Data and AI
Google Pixel Phone -“Best Take”
iPhone -Classify people & pets!
https://blog.google/products/photos/how-google-photos-best-take-works/
https://support.apple.com/en-gb/guide/iphone/iph9c7ee918c/ios
Computer Vision

56. 2025
The Peak of Data and AI
Computer Vision
●If AI enables computers to think, computer vision
enables them to see!
●OCR –An early computer vision, developed in 1974
●That early tech now gives us real-time translation

57. 2025
The Peak of Data and AI
Computer Vision
●You have clever optics in your head
●But, computers have needed to be trained
●…and, fun fact… you’ve likely helped!

58. 2025
The Peak of Data and AI
CAPTCHA
“completely automated public Turing test to tell computers and humans apart.”

59. 2025
The Peak of Data and AI
Demo –Vision Comparison

60. 2025
The Peak of Data and AI

61. 2025
The Peak of Data and AI
Prompt engineering
https://www.kaggle.com/whitepaper-prompt-engineering

62. Workspace Group –Case Study
https://l.ead.me/fmecasestudy

63. FME Hub –Verified Publisher

64. Computer Vision -R&D
●Utilisingpre-trained models -Machine Learning
https://universe.roboflow.com/

65. 2025
The Peak of Data and AI
Demo –Orchestration

66. 2025
The Peak of Data and AI

67. 2025
The Peak of Data and AI

68. 2025
The Peak of Data and AI
UKPN –The Challenge
Identify the Cross-Section line and associated number from a map.
Review the ground layout for digitisation.

69. 2025
The Peak of Data and AI
UKPN –e2e Solution

70. 2025
The Peak of Data and AI
UKPN –e2e Solution (output)
Roboflow, Train the model: 1 day
Model Generation (Cloud Computing): 6-8 hrs
Object, Text Detection Success Rate: 75%-80%
Cost:
Roboflow$250 per month
Gemini $4

71. 2025
The Peak of Data and AI
Simon Green
Principal Technical Specialist
Siobhan Ryan
Junior Technical Specialist
Oliver Morris
Business Directorhttps://fme.safe.com/webinars/

72. 2025
The Peak of Data and AI
https://www.bbc.co.uk/programmes/m001wjf8

73. 2025
The Peak of Data and AI
“…and finally, a bit of fun!”

74. 2025
The Peak of Data and AI
Grab your phones!
Do:
●Launch the App
●Tap ‘browse file system’
●Snap something that belongs to you…
●Tap ‘Run’ and tell us how you were
categorised!!
Don’t:
●Take a photo of yourself or someone
else
●Upload a photo showing personal data
https://l.ead.me/FMEmyself

75. 2025
The Peak of Data and AI
Grab your phones!
Do:
●Launch the App
●Tap ‘browse file system’
●Snap something that belongs to you…
●Tap ‘Run’ and tell us how you were
categorised!!
Don’t:
●Take a photo of yourself or someone
else
●Upload a photo showing personal data
https://l.ead.me/FMEmyself

76. 2025
The Peak of Data and AI
Thank You
David Eagle
Avineon Tensing
[email protected]