Know AI
The Three V's in Big Data are
1 Volume: The amount of Data generated 2. Variety. The heterogeneity of the data 3. The rate of change in said data
Steps for closing with a client
1. Close with a suggested topic for the next meeting 2. Follow up with an email providing resources 3. Include the Booz Allen AI platform as necessary in follow-up discussions
5 Core elements to building an AI solution
1. Data 2. Algos 3. Training 4. Eval 5. Deployment
AIOps is broken down into thre sub ops
1. Dataops 2. MLOPS 3. DevSecOPS
Reinforcement learning broken down in steps
1. Establish a goal 2. Initialize the learning process 3. Agent 4. Perform actions 5. Env feedback
Our AI Strategy for acquisition is broken
1. Hypergrowth, 2. AI Fabrication 3. Exceptional Quality 4. Extraordinary Talent
How will boozy keep continuting to win the market
1. Minmizing time cost and risk 2. Bringin our depth of talent 3. Deploying out understanding of Data, software and security architechtures 4. Promoting an apply-first philosophy 5. Continuing to embrace a "mission-first" focus with client
Unsupervised Learning
1. No labels are added 2. Look and learn to see what patterns are made on based on the model 3. Used for clustering or anomaly detection
Supervised learing steps
1. Putting a label on times into the model. 2. Feeding machine unlabeled data to see if it can distinguish the two 3. Refine if needed. Used for classification or regression
The two major branches for the development of an AI project are
1. The Business Analytics 2. Model Development
Boozy is having and AI first approach
1. Through Mainaining industry leadership 2. Avoiding Risk (Moving quickly, Embracing new technologies, )
Cyber Precog
A GPU unit that processes data many times faster. Than the CPU A cyber precog is a term that can refer to an individual or an artificial intelligence system that predicts cyber threats, attacks, or security incidents before they happen.
DevSecOps (Development, Security and Operations)
A combination of software development, security operations, and systems operations by integrating each discipline with the others
Deep learning
A specialized collection of machine learning appraches leverating dep neural networks to enable AI capabilities
Page 237 for the resources pick oe and review it
AI resources sheet
Client Engagement resources
Ai Primer AI calling card deck AI sell Deck Booz Allen external site Boozy Internal site AI yammer AI teams point of contact
Boozy takes an agressive pathway
Always asking why not AI Moving with speed and velicity and taking an advantgage with our AI position
aiSSEMBLE what is it
Booz Allens lean manufactuaing approach to designing developing and fielding AI solutions accross the engineering lifecycle
Boozy Provides AI engineering
By using AI Engineering platforms Data Modeling and pipelines AI enabled systems and platforms AI testing and evaluation
GPU (graphics processing unit)
Can not only render images but can handle massive parralel computations which has contributed to AI
Naive Bayes
Classification predictive Training data classify new data points ie red, round, Apple, yellow, oblong,banana If new obj red then more likely Apple
Three types of clients
Clients starting on their journey Clients that have started the journey Clients that are fully on their way to an AI future
Contextual reasoning AI
Closer to human intelligence AI, Can self improve with data and can operate in dynamic envs
Boozy Provides AI model development
Computer Vision Language Models Multamodal Models Counter AI Generative AI Reinfirment Models and Autonomy
The three core roles in AI are
Data Scientist, ML Engineer Data Engineer
DataOps is the process
Data management. Where you gotta consider Accuracy Usability Provenance and Security
The three major issues for AI solutions at the enterprise level are
Deployment Assurance Scaling
Synthetaic
Enables dev, data scientits and analyts to quickly developm object detection algorithims for rare objects without the need for labeled data
ANI (Artificial Narrow Intelligence)
Executes specific tasks well
HCAI
Human centerd AI- A combo of ethics and AI
Pattern Recognition AI needs
Massive Data, Data must be tagged labeled and organized Resource intensive
BigHorn
Mobile AI development kit that allows AI solutions where there is low accessibility to the cloud such as in Battle Areas
PWIN
Probability of winning a gov contract
Boozy techniques whn we propose our services
Say the we minimze time and cost to impact We build sustainable solutions We drive adoption
AGI (Artificial General Intelligence)
Strong AI allows machines to achienve human-level cognitive abilities
Some ML algorithim types are
Supervised Learning, Unsepervise Learning, Reinforcement Learning, and Neural Networks
Precision formula
TP/(TP+FP)
Neural networks
There is the input layer which are Neurons. Hidden Layers, with weights,. that make complex abstractions the outpul later with put a label on whatever the input is
Boozy provides AI strategy and adoption
Through, Safe and Ethical Implementation 2. Through Data science and research 3. Through Workforce tranind 4. Thorugh scouting the best technology
Latent AI
TinyMl develper tools enable Booz Allen to deploy algorithms in low compute, low-power envs
AI Systems should be
Transparent Accountable Explainable Privacy Enhanced Fair with Harmful Biased managed
Three factors in AI advancement
Vast amounts of Data, Growth in computin power, Advances in deep learning research
Data comes in many forms
You have: 1. Structured Data 2. Multimodal Data 3. Labeled Data 4.Unlabeled Data
Ai.bah.com
check it out
We can increase our PWIN
if we present AI as an option
We'll provide the AI systems
in the cloud in the data center, portable or on device
aiSSEMBLE
our Boozy AI portfolaio
SWaP
security, size, weight, and power
AIOps
the process, strategies and framworks to operatinalize AI to address real-world challenges
R squared
the proportion of the total variation in a dependent variable explained by an independent variable
If we are in RFP
we can still provide AI as a feature
While on a project
we can still provide AI as an adoption
