Accelerating Outcomes in Big Data, IIoT/IoT, and AI/ML Talking AI - - PowerPoint PPT Presentation
Accelerating Outcomes in Big Data, IIoT/IoT, and AI/ML Talking AI with Hashmap July 2018 2 HASHMAP : WHAT WE DO We accelerate innovative business outcomes in BIG DATA, IIOT/IOT, and AI/ML with a combination of PEOPLE, PROCESS, AND TECHNOLOGY
Talking AI with Hashmap
July 2018
2
We accelerate innovative business outcomes in BIG DATA, IIOT/IOT, and AI/ML with a combination of PEOPLE, PROCESS, AND TECHNOLOGY enabling END-TO-END SOLUTION DESIGN, DEVELOPMENT, & DEPLOYMENT
HASHMAP : WHAT WE DO
3
Hashmap Corporate Snapshot
–Founded in 2012 to provide Big Data Consulting –HQ in USA (Roswell, GA) with offices in Houston, Canada, and India –Sole focus has always been Big Data & IIoT/IoT –Business model is 100% consulting services and software engineering services (we do NOT resell hardware, software, or subscriptions) –Subcontract our services - you make $$$
HASHMAP : ABOUT US
HASHMAP QUICK STATS
ATLANTA 1000 Holcomb Woods Parkway Building 100, Suite 118 Roswell, GA 30076 HOUSTON 24275 Katy Freeway Suite 400 Katy, TX 77494 INDIA Midas Tower, Plot # 44, RGIP Phase 1, Hinjewadi, Pune, Maharashtra, India
hashmapinc.com
Consulting Services
Strategy, Architecture, Data Engineering, Solution Consulting, and Application Development
OUR SERVICES FOCUS AREAS
Accelerator Services
template-based approach
Managed Services
Models
Optimization, Automation, Improvement
CANADA 4145 North Service Road 2nd floor Burlington, Ontario L7L6A3
HASHMAP : ABOUT US
4
CONSULTING PROJECTS COMBINED CUSTOMERS INDUSTRIES
Oil & Gas Financial Services Technology Manufacturing Insurance Retail Pharma Healthcare Communications Power & Utilities Digital Media
ROW
1
PRESENTED BY DATABRICKS AT SPARK - AI SUMMIT 2018
5
Only a small % of enterprises are successful with AI & ML
PRESENTED BY DATABRICKS AT SPARK - AI SUMMIT 2018
6
Everyone Else Faces Major Challenges That Slow Down AI/ML Projects
DATA IS NOT READY FOR ANALYTICS A ZOO OF AI/ML FRAMEWORKS IT IS HARD TO PRODUCTIONIZE AI/ML
7
IIoT/IoT and Streaming Analytics
– Tempus IIoT/IoT Framework – Industrial control systems integration – Time Series Dataset Optimization – Edge Intelligence and Edge Computing – Deep expertise in Kafka and Spark Streaming – Data flow and data
HASHMAP : CONSULTING
NoSQL
& SQL
– Evaluation, comparison, use case-based implementation – Optimization, Tuning, Performance Testing, and Benchmarking – Automated data ingestion into Big Data platforms for SQL datasets – Low Latency Operational Data Stores
Cloud and Platform Services
– Container Orchestration and Management – DevOps – Cloud Enablement and Simplified Deployment – Data Virtualization – Security – Governance – Operations – Performance
Data Science, Analytics, AI, & ML
– Rapid Analytics Application Creation – Self Service ML – In Memory Analytics Grid – Discovery & Exploration – Data insights – Data Preparation – Predictive Analytics
Data Engineering &
Data Pipelines
– Data Integration – Data Quality and Curation – Spark Data Pipelines – Data Warehousing, EDW Enablement, Lake Shore Marts – Single View and Customer 360 – Offloads and Data Archiving – Enterprise Application Environments – Batch and Real Time Ingestion
8
HASHMAP : PROJECTS
POC to determine marketing content effectiveness using Natural Language Processing ”NLP” in order to better focus ongoing marketing spend and budget
CHALLENGE
Automated social feed ad content for data acquisition, data cleansing and curation, natural language processing, visualization and data export to provide content analysis, engagement analysis, and spend analysis
APPROACH
Provided quick visibility into marketing content effectiveness through engagement prediction, post clustering, and key word identification associated with engagement success
OUTCOME
9
HASHMAP : PROJECTS
Integration of on-premise, real time streaming data sources including an enterprise historian with dynamic, predictive ML models to provide insights to field operations
CHALLENGE
Integrated real time data collection and a high speed message broker with the
providing a dynamic streaming solution for automated ML pipelines
APPROACH
A production ready, secure solution allowing the data science team to quickly
productionize their ML models within hours
OUTCOME
1
HASHMAP : OUTCOME & USE CASE ENABLEMENT
HASHMAP OUTCOME & USE CASE ENABLEMENT WORKSHOP
Attendees
Program Key Stakeholders (Business and IT)
Duration
1-2 days onsite, 1-2 days offsite
Onsite Agenda
⏤ Business Context – Detailed use case review and success criteria ⏤ Architecture Review ⏤ Data Center / Cloud Review ⏤ Ongoing Operations ⏤ Program/Project Time Horizons and Enabling Roadmap ⏤ Confirmation / Final Risk Identification / Deliverables Discussion / Wrap up
Deliverables
⏤ Business Value Matrix and Detail for Specific Use Cases ⏤ Conceptual Architecture (Edge/Facility and Data Center/Cloud) and program/project horizon mapping ⏤ Recommendations and Go-Forward Plan
Key Dimensions Reviewed in the Business Context Segment
⏤ Organizational Effect ⏤ Business Impact ⏤ Priority and Urgency ⏤ Expected Time to Deliver ⏤ Risks ⏤ Innovation or Cost Savings Focus and Impact ⏤ User Population and Consumption Patterns ⏤ Capability Assessment and Review ⏤ Datasets ⏤ Other Dependencies ⏤ Technologies/Tools in use/planned/vision ⏤ ML/AI requirements ⏤ Analytics and visualization requirements
UNDERSTANDING – COLLABORATION - STRATEGY – ROADMAP - ARCHITECTURE
Objectives
DESIGN, REFINE, DEVELOP, DELIVER with an
Facilitate discussion and documentation of… ⏤ Business Value ⏤ Solution Architecture ⏤ Ongoing Operations ⏤ Timeline/Horizons Provide deliverables that… ⏤ Demonstrate business alignment ⏤ Outline initiatives, priorities, and time horizons ⏤ Highlight expected value
HASHMAP : OUTCOME & USE CASE ENABLEMENT
11
12
Review and Field Testing
HASHMAP : LABS
Edge
Intelligence
Blockchain
Enablement
Industrial
Deep Learning
Computer
Vision
Container
Orchestration
Self Service
ML
13
HASHMAP : LABS
In hazardous areas such as an Oilfield Drilling Rig, a Manufacturing Plant, or Heavy Equipment facility, it is important to understand the exposure of personnel to the riskiest areas. Gathering this data is crucial in understanding the exposure for improvements in operational processes and risk mitigation.
CHALLENGE
Using a convolutional neural network (image classification algorithm) applied to the whole image at a single point in time, several frames can be analyzed per second. By splitting a video image down into representative frames in real time at the edge, we are able to transmit only the events, rather than an entire image or video segment.
APPROACH
By applying real-time object detection, risk assessments and mitigation can be undertaken before the occurrence of a significant safety incident. By including safety related KPIs in a performance tracking program, organizations can have a holistic performance target in terms of both efficiency and safety.
OUTCOME
Utilizing a small edge device such as an Nvidia Jetson SoC (system on a chip) or a a ruggedized device, running a pre-trained
identification of objects of interest and the confidence of the identification. Based on a configurable threshold, events can be generated and transmitted over MQTT/CoAP
SOLUTION
Contact Hashmap directly to discuss
Kelly Kohlleffel VP Sales & Marketing (713) 628-6030 kelly@hashmapinc.com linkedin.com/in/kellykohlleffel
Check out our Engineering & Technology Blog
https://medium.com/hashmapinc
Subscribe to our IoT on Tap Podcast
https://www.hashmapinc.com/podcasts