Use this quick start guide to collect all the information about UiPath UiSAI Certification exam. This study guide provides a list of objectives and resources that will help you prepare for items on the UiPath Specialized AI (UiSAI) exam. The Sample Questions will help you identify the type and difficulty level of the questions and the Practice Exams will make you familiar with the format and environment of an exam. You should refer this guide carefully before attempting your actual UiPath Certified Professional Specialized AI (UiSAI) certification exam.
The UiPath UiSAI certification is mainly targeted to those candidates who want to build their career in Specialized AI domain. The UiPath Certified Professional Specialized AI (UiSAI) exam verifies that the candidate possesses the fundamental knowledge and proven skills in the area of UiPath UiSAI.
UiPath UiSAI Exam Summary:
| Exam Name | UiPath Certified Professional Specialized AI (UiSAI) |
| Exam Code | UiSAI |
| Exam Price | $150 (USD) |
| Duration | 180 mins |
| Number of Questions | 60 |
| Passing Score | 70% |
| Books / Training |
Automation Developer - Specialized AI Training Specialized AI Professional OLT |
| Schedule Exam | Pearson VUE |
| Sample Questions | UiPath UiSAI Sample Questions |
| Practice Exam | UiPath UiSAI Certification Practice Exam |
UiPath UiSAI Exam Syllabus Topics:
| Topic | Details |
|---|---|
| UiPath Document Understanding |
- Define what UiPath Document Understanding is. - Distinguish between structured, unstructured, and semi-structured documents that can be processed with Document Understanding. - Differentiate between the two common types of data extraction methodologies: rule-based and model-based. - Differentiate between OCR and Document Understanding. |
| UiPath Document Understanding Framework |
- Describe the stages of the Document Understanding Framework. - Use the Document Understanding Framework to build document processing workflows in UiPath Studio. |
| UiPath Studio - Document Understanding Activities |
- Select and install the Studio Intelligent OCR package activities used for information extraction. - Use the Taxonomy Manager to create a taxonomy for the project. - Load the taxonomy in a variable to be used in conjunction with other activities. - Digitize scanned or digital documents using the Digitize Document activity. - Explain what a Document Object Model is in the context of Document Understanding. - Select an appropriate OCR engine for your digitization use case. - Analyze what kind of classifier/extractor is the most suitable for the automation project. - Classify the scope of documents using Classify Document Scope activity and the Classifiers Wizard. - Extract data from the documents using relevant Studio Activities, extractors, and Configure Extractors Wizard. - Distinguish between different types of extractors based on the structure of documents. - Select and configure pre-trained or newly trained Machine Learning extractor for data extraction. - Use the Classification/Validation Station to review and correct document classification and automatic data extraction results. - Use UiPath Orchestrator or Action Apps to configure human validation steps. - Define how to export data. - Export Extraction Results using relevant Studio Activities. - Train the Classifiers and Extractors used to improve their performance. |
| DU Specific UiPath Implementation Methodology |
- Gather and analyze data about the documents in scope (document types, fields extracted, pages per document). - Gather and analyze data about languages in scope and OCR engine of choice. - Integrate exception handling within the automation solution. - Describe best practices when building a ML model. - Describe the selection of a model. - Describe best practices for process design. - Describe the metering or charging of Page Units and AI Units |
| UiPath AI Center |
- Define what UiPath AI Center is. - Explain how RPA and AI can work together in process improvement. - Distinguish between AI, ML, NLP, DL, Computer vision. - Describe how machine learning works. - Distinguish between supervised, unsupervised and reinforcement learning. - List the applications of machine learning across different industries. - Describe how AI Center works. - List the user personas who can access and use AI Center. - List the types of ML models in AI Center. - Describe the various ways to deploy and install AI Center. - List the example of out-of-the box ML Packages application. - Define the AI Center User Interface elements. - Manage projects in AI Center (Create, edit, delete). - Manage datasets in AI Center (Create, upload, edit, delete, make the dataset public). - Manage data labels in AI Center (Create a data labeling instance, configure). Build ML Packages in AI Center. - Manage ML Packages in AI Center (Upload ML Package, import, view ML Package details, version control of ML Packages). - Use Out-of-the-box ML Packages from AI Center. - Manage the Pipelines available in AI Center (Create, schedule pipelines, edit scheduled pipelines, remove pipelines). - Describe how to retrain a model by sending feedback from the process to the model. - Create ML skills. - Update ML Skills in new ML Packages (Upload ML Package, import, view ML Package details, version control of ML Packages). - Describe the steps to make an ML skill public. - Describe the types of events captured in the ML logs. - Build the first model and test it out for labeling and types. - Evaluate the model. |
| UiPath Document Understanding Process Template |
- Describe what the DU Process template is. - Explain the architecture of the DU Process. - Explain the exception handling within the automation solution. - Explain Classification/Extraction Validation. - Explain the Post Processing implementation. - Explain the Locking mechanisms in the DU Process. - Explain the Test implementation in the DU Process. - Explain the importance of Export/End Process. - Explain the best practices regarding Data Export. - Explain the Config file. - Explain Main-Action Center & Main-Attended. |
| UiPath Communications Mining |
- Distinguish between Communications Mining, Process Mining and Task Mining. - Describe how UiPath Communications Mining works. - Distinguish between the two most common scopes of using UiPath Communications Mining (Analytics, Automation). - Distinguish between optimal and sub-optimal data types that Communications Mining can interpret and add structure to. - Explain how UiPath Communications Mining and RPA can work together (including DU). - Describe the Communications Mining user interface elements (admin and project views). |
| UiPath Communications Mining - Model Training |
- Explain, at a minimum, what describes a high performing model. - Explain verbatims, labels, entities, metadata, and how to use them. - List and correctly order the stages of the Communications Mining Process. - Describe the golden rules of label training. - Describe the golden rules for entity training. |
| UiPath Communications Mining - Taxonomy Design |
- Describe the 'Taxonomy Design' phase of the model training process. - Create a label taxonomy structure according to best practices. - Differentiate between analytics and automation taxonomies. - Provide examples of typical groups of labels (process/request types, quality of service / failure demand etc.). - Distinguish between different types of entities (pre-trained, trained from scratch, trainable, non-trainable). |
| UiPath Communications Mining – Setup |
- Describe the three main components of data (data sources, datasets, projects) and how to manage them. - Enable, update, or disable entities in a dataset in UiPath Communications. - Import a taxonomy via the Settings or Train pages in UiPath Communications Mining. - Distinguish between tone analysis and label sentiment. |
| UiPath Communications Mining – Discover |
- Label clusters considering key best practices. - Describe what the Search functionality in Discover is and when it is recommended to use it. - Explain the risks associated with using too much Search to train the model vs balancing out with Shuffle and Teach Label. |
| UiPath Communications Mining – Explore |
- Explain what label and entity predictions are, how they work, and how to use them. - Distinguish between label predictions and label suggestions. - Differentiate between when it makes sense to use Shuffle, Teach Label, or Low Confidence to train the model in the 'Explore' phase. - Use Shuffle, Teach Label, or Low Confidence in the 'Explore' phase according to best practices. - Explain when it is recommended to use Teach Entity to continue label training at the end of the Explore phase. - Prune and reorganize a taxonomy by editing, renaming, merging and deleting labels. - Prune and reorganize a taxonomy by modifying or deleting an entity. |
| UiPath Communications Mining - Refine and Maintain |
- Define why the 'Refine' phase of the Communications Mining process is important. - List the steps to be performed in the 'Refine' phase of the model training process. - Explain precision and recall metrics, how they impact the performance of machine learning models. - Describe what Model Rating assesses, and what factors it takes into consideration (Performance, Coverage, Balance). - Analyze All Labels and suggest typical solutions to improve the score (understand MAP). - Analyze Underperforming Labels and suggest typical solutions to improve the score. - Analyze Coverage to check how well covered the whole dataset is and suggest typical solutions to improve the score. - Analyze Balance to check for a balanced representation of the whole dataset and suggest typical solutions to improve the score. |
| Distinguish between the three label performance indicators (blue, amber, red) |
- List potential reasons that can lead to low label performance. - Address bias labelling by continuing to train the model using Teach Label in the 'Refine' phase of the Communications Mining process. - Continue to train the model using Check Label and Missed Label in the 'Refine' phase of the Communications Mining process. - Name the recommended Model Ratings for automation and analytics use cases. - List indicators of when model training could be finished at the end of the UiPath Communications Mining process. - Analyze what entity scores are, how they are calculated, and typical solutions to improve them (Teach Entity, Check Entity, Missed Entity). - List the two key factors that can erode a model's performance (brand new labels are added, but not trained, or concept drift occurs.) - Explain how to add new labels to an existing taxonomy. - Explain how models should be maintained in production. |
| Analytics & Monitoring |
- Use the 'Reports' pages to create customized and dynamic dashboards. - Use the 'Label Summary' tab to analyze charts and high-level summary statistics. - Use the 'Trends' tab to analyze trends for verbatim volume, label volume, sentiment over a given time period, etc. - Use the 'Segments' tab to analyze label volumes versus verbatim metadata fields, e.g., Sender Domain. - Use the 'Comparison' tab to conduct A/B tests and cohort comparisons between different cohorts of the dataset. - Use the 'Threads' tab to analyze conversations and their characteristics. - Use 'Quality of Service' and 'Tone Analysis' to monitor channel performance. - Use 'Alert Center' to configure and track alerts and issues. |
| Automation and Model Management |
- Apply CI/CD best practices for model management. - View, create and modify streams. - Choosing the right thresholds for streams. - Pinning model versions for productions and staging. - Describe the Knowledge Ingestion Framework. |
To ensure success in UiPath UiSAI certification exam, we recommend authorized training course, practice test and hands-on experience to prepare for UiPath Specialized AI (UiSAI) exam.