Atlassian

Preparation Guide

Preparation Strategy

1. Focus on ML Infrastructure: Discuss various aspects of building and maintaining ML systems at scale. This includes topics like:

Note

• Distributed training architectures

• Model serving and deployment strategies

• Data pipeline design and management

• MLOps practices

2. Deep Learning Optimization: Be prepared to discuss:

Note

• Different attention mechanisms (self-attention, cross-attention, multi-head attention)

• Hardware bottlenecks in training and inference (memory bandwidth, compute limitations)

• Optimization strategies for large language models

• Best practices for Supervised Fine-Tuning (SFT), Proximal Policy Optimization (PPO), and Direct Preference Optimization (DPO)

• Advanced techniques like Flash Attention and Paged Attention

• Quantization techniques and their trade-offs

3. AI Acceleration: Be prepared to discuss strategies for rapid AI adoption and integration within an enterprise setting.

4. Scalability and Performance: Be ready to discuss how to design systems that can handle large amounts of data and users efficiently.

Depth of technical discussion:

While the ML design round will likely focus on solving a particular business problem, given the job description, you should be prepared to go into significant technical depth if asked. However, remember to balance this with high-level system design and business considerations.

Here’s how I suggest you approach the interview:

1. Start with the high-level business problem and system design.

2. As you explain your solution, be prepared to dive deep into technical details, especially around ML infrastructure and optimization techniques.

3. Don’t proactively go into extreme technical depth unless asked, but have that knowledge ready to demonstrate when prompted.

4. Always tie your technical decisions back to business impact and user experience.

The interviewer will be evaluating not just your technical knowledge, but also your ability to communicate complex ideas clearly and your understanding of how technical decisions impact business outcomes.

Be prepared to discuss the implementation details of LLMs, including attention mechanisms, optimization strategies, and advanced techniques like flash attention and quantization. However, don’t force these topics into the conversation if they’re not directly relevant to the problem at hand. Instead, use them to demonstrate depth of knowledge when appropriate or when specifically asked about them.

Key Areas

Warning

• Detailed understanding of fine-tuning processes, especially LoRA and other parameter-efficient methods.

• Practical considerations for deploying LLMs in enterprise environments (like Atlassian’s products).

• Techniques for improving efficiency and reducing computational costs in LLM applications.

• Methods for ensuring data privacy and security when using LLMs with potentially sensitive enterprise data.

• Strategies for continual learning and model updates in production environments.

See also

1. Understand the Problem: Start by asking clarifying questions to fully understand the problem during the interview.

2. Communicate Clearly: Explain your thought process and the rationale behind each decision you make.

3. Draw from Experience: Relate your design to past projects and experiences.

4. Focus on Training Process: Spend time detailing the training process, including data preparation, model training, and fine-tuning techniques.

5. Practice Design Problems: Practice designing systems on a whiteboard or paper to simulate the interview environment.

LLM Application Area

1. Document Understanding: Document classification and information retrieval, Text summarization

2. Question Answering Systems: Advanced methods for generating accurate and contextually relevant answers.

3. Text Generation: Chatbots and conversational AI

4. Code Generation: Code generation and completion

Atlassian Products Using This Tech

1. Confluence: Enhanced search and Q&A capabilities

2. Jira: Automated ticket classification and routing

3. Trello: Natural language task creation and management

4. Bitbucket: Code review assistance and documentation generation

Things You Must Know About This Tech

1. Transformer architecture

2. Pre-training and fine-tuning techniques

3. Prompt engineering

4. Few-shot and zero-shot learning

5. Retrieval-Augmented Generation (RAG)

6. Parameter-efficient fine-tuning methods (e.g., LoRA, Adapters)

ML Theory You Must Know

1. Attention mechanisms

2. Transfer learning

3. Tokenization strategies

4. Embedding techniques

5. Overfitting and regularization in large language models

6. Optimization algorithms for large-scale training

Trade-offs in Different Modeling Choices

1. Model size vs. inference speed

2. Fine-tuning vs. prompt engineering

3. Generative vs. discriminative approaches

4. Open-source vs. proprietary models

5. On-premise vs. cloud deployment

Metrics and Evaluation

1. Perplexity: Common metric for language models to measure uncertainty.

2. BLEU, ROUGE, METEOR: for text generation

3. Precision, Recall, F1-Score: For tasks like classification and information retrieval.

4. Human evaluation metrics (e.g., coherence, relevance)

5. Bias and fairness metrics

MUST KNOW Research Papers in the Past 3 Years

1. Attention is All You Need

2. BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding

3. T5: Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer

4. GPT-3: Language Models are Few-Shot Learners

5. InstructGPT: Training language models to follow instructions with human feedback

6. LaMDA: Language Models for Dialog Applications

7. PaLM: Scaling Language Modeling with Pathways

8. FLAN: Few-Shot Learning with Task Descriptions

9. Low-Rank Adaptation of Large Language Models

Sample ML Problems

Design a Content Recommendation System for enhancing knowledge discovery in Confluence Cloud

Problem Statement:

Confluence Cloud serves as a central repository for documentation, wikis, and collaborative content creation within organizations. However, users often struggle to discover relevant content amidst the vast amount of information stored in Confluence pages. Design a content recommendation system that leverages machine learning algorithms to analyze user behavior, content attributes, and collaboration patterns, in order to provide personalized recommendations for knowledge discovery and exploration within Confluence Cloud.

Key Insights and Signals:

1. Understanding of Confluence Cloud’s Content Ecosystem:

• Does the candidate have a comprehensive understanding of how content is structured and organized within Confluence Cloud?

• Can they identify common challenges faced by users in navigating, searching, and accessing relevant information within Confluence pages?

2. User Behavior Analysis:

• How does the candidate propose to capture and analyze user interactions, content views, and collaboration activities within Confluence Cloud?

• Are they familiar with techniques such as user profiling, session tracking, and content affinity modeling for extracting meaningful insights from user data?

3. Content Attributes and Metadata Extraction:

• What features does the candidate suggest extracting from Confluence pages, including titles, tags, labels, and attachments, to characterize content attributes?

• Can they incorporate contextual information such as page categories, author expertise, and viewer preferences to improve recommendation relevance?

4. Recommendation Algorithms:

• Does the candidate propose algorithms for generating personalized content recommendations based on user interests, content relevance, and collaborative filtering?

• Are they able to balance between popularity-based recommendations and more personalized approaches to cater to diverse user preferences?

5. Integration with Confluence Cloud Interface:

• How does the candidate plan to integrate the content recommendation system seamlessly into the Confluence Cloud user interface and search functionality?

• Can they propose widgets, plugins, or search extensions for delivering recommendations directly within Confluence pages and search results?

6. Evaluation and Feedback Loop:

• What metrics does the candidate suggest for evaluating the effectiveness and utility of the content recommendation system in facilitating knowledge discovery?

• Are they able to incorporate mechanisms for collecting user feedback, measuring recommendation relevance, and iteratively refining the recommendation algorithms based on user engagement metrics?

Rating Criteria:

• Content Understanding (5/5): Demonstrates deep knowledge of Confluence Cloud’s content ecosystem, user needs, and information retrieval challenges.

• Personalization Techniques (4/5): Provides innovative approaches for generating personalized content recommendations tailored to individual users and usage contexts.

• Integration and Usability (4/5): Addresses technical challenges in integrating the recommendation system with Confluence Cloud’s architecture and UI for seamless interaction and exploration.

• User-Centric Design (4/5): Considers usability, relevance, and contextualization of recommendations to enhance user experience and knowledge discovery in Confluence Cloud.

• Impact Assessment (4/5): Identifies potential benefits of the content recommendation system in terms of improved content discoverability, collaboration, and productivity within Confluence Cloud.

Intelligent Q&A System for improving knowledge sharing in Confluence Cloud

Problem Statement:

Confluence Cloud, Atlassian’s collaboration software, serves as a central knowledge base for teams to document and share information. However, users often face challenges in finding relevant answers to their questions buried within Confluence pages. Design an intelligent Q&A system that leverages natural language processing (NLP) and machine learning (ML) techniques to enhance knowledge discovery and facilitate seamless information retrieval within Confluence Cloud.

Key Insights and Signals:

1. Understanding of Confluence Cloud’s Use Cases:

• Does the candidate have a clear understanding of how Confluence Cloud is used for documentation, knowledge sharing, and collaboration?

• Can they identify common scenarios where users seek answers to questions within Confluence?

2. Natural Language Processing (NLP):

• How does the candidate propose to extract and analyze textual content from Confluence pages to understand the semantics and context of user queries?

• Are they familiar with NLP techniques such as named entity recognition, sentiment analysis, and topic modeling for processing unstructured text data?

3. Question Understanding and Intent Recognition:

• What methods does the candidate suggest for interpreting user questions and identifying the underlying intent or information needs?

• Can they propose algorithms for query expansion, disambiguation, and entity linking to improve the accuracy of question understanding?

4. Knowledge Graph Representation:

• Does the candidate address the challenge of representing Confluence content as a structured knowledge graph to capture relationships between topics, documents, and concepts?

• Are they able to propose techniques for entity extraction, entity linking, and knowledge graph construction from unstructured text data?

5. Semantic Search and Relevance Ranking:

• How does the candidate plan to implement semantic search algorithms that leverage the knowledge graph to retrieve relevant answers to user questions?

• Can they incorporate techniques such as semantic similarity, graph-based ranking, and context-aware search to improve result quality?

6. Integration with Confluence Cloud Interface:

• How does the candidate propose to integrate the intelligent Q&A system seamlessly into the Confluence Cloud user interface?

• Can they ensure that the Q&A functionality is intuitive, accessible, and closely integrated with existing Confluence features?

Rating Criteria:

• NLP Expertise (5/5): Demonstrates proficiency in NLP techniques and their application to text analysis and understanding.

• Semantic Understanding (4/5): Provides innovative approaches for representing and querying knowledge in Confluence Cloud using semantic technologies.

• User-Centric Design (4/5): Considers usability, relevance, and accessibility of the Q&A system to enhance user experience and knowledge sharing.

• Integration and Interoperability (4/5): Addresses technical challenges in integrating the Q&A system with Confluence Cloud’s architecture and APIs.

• Impact Assessment (4/5): Identifies potential benefits of the intelligent Q&A system in terms of improved knowledge discovery, collaboration, and productivity within Confluence Cloud.

Enhance the search and recommendation features in Jira Cloud

Problem Statement:

Jira Cloud, Atlassian’s flagship product for agile project management, aims to improve user productivity and collaboration. One common pain point reported by users is the challenge of finding relevant information quickly and receiving personalized recommendations for tasks and workflows within Jira. Design a system using Large Language Models (LLMs) to address these issues and enhance the search and recommendation capabilities of Jira Cloud.

Key Insights and Signals:

1. Understanding of Jira Cloud’s Functionality:

• Does the candidate have a clear understanding of the features and workflows within Jira Cloud?

• Can they identify specific use cases where improved search and recommendation capabilities would benefit users?

2. Domain-specific Knowledge:

• Does the candidate demonstrate familiarity with agile project management concepts and terminology?

• Are they able to tailor the LLM-based solution to the unique requirements of Jira Cloud users?

3. User Intent Recognition:

• How does the candidate propose to interpret user queries and understand their intent within the context of Jira tasks and projects?

• Can they suggest techniques for semantic understanding and contextual relevance in search results and recommendations?

4. Personalization and Contextualization:

• Does the candidate address the challenge of providing personalized recommendations based on user preferences, project history, and collaboration patterns?

• Are they able to incorporate contextual information such as project metadata, user roles, and task dependencies to improve recommendation accuracy?

5. Integration with Jira Cloud Infrastructure:

• How does the candidate plan to integrate the LLM-based search and recommendation system seamlessly into the Jira Cloud platform?

• Can they propose APIs, webhooks, or other integration mechanisms to ensure interoperability with existing features and workflows?

6. Performance and Scalability:

• What measures does the candidate suggest for optimizing the performance and scalability of the LLM-based system within the Jira Cloud environment?

• Are they able to balance computational resource constraints with real-time responsiveness and user experience?

Rating Criteria:

• Domain Expertise (5/5): Demonstrates in-depth knowledge of Jira Cloud’s functionalities and user needs.

• Customization and Personalization (4/5): Provides innovative solutions for tailoring search and recommendations to individual user contexts.

• Technical Feasibility (4/5): Proposes realistic approaches for integrating LLM technology into Jira Cloud’s infrastructure.

• User-Centric Design (4/5): Considers usability, relevance, and user feedback mechanisms in the design process.

• Business Impact (4/5): Identifies potential benefits of the proposed solution in terms of user satisfaction, productivity gains, and competitive advantage for Atlassian.

Design an Intelligent Chatbot for improving customer support in Jira Service Management

Problem Statement:

Jira Service Management, Atlassian’s service desk solution, is used by organizations to manage IT service requests, incidents, and support tickets. However, users often experience delays and inefficiencies in resolving issues due to long response times and repetitive queries. Design an intelligent chatbot powered by natural language processing (NLP) and machine learning (ML) techniques to provide proactive assistance, automate routine tasks, and streamline customer support interactions within Jira Service Management.

Key Insights and Signals:

1. Understanding of Jira Service Management Workflow:

• Does the candidate have a clear understanding of how Jira Service Management is used for managing service requests and incidents?

• Can they identify common pain points in the customer support workflow, such as ticket triaging, issue resolution, and communication with end-users?

2. Natural Language Understanding (NLU):

• How does the candidate propose to interpret user queries and extract relevant information from support tickets and service requests?

• Are they familiar with NLP techniques such as intent classification, entity recognition, and sentiment analysis for understanding user intent and context?

3. Automated Ticket Triage and Routing:

• What methods does the candidate suggest for automating ticket triaging and routing based on the content and urgency of support requests?

• Can they propose algorithms for classifying tickets, assigning priority levels, and escalating critical issues to appropriate support teams?

4. Contextual Assistance and Knowledge Retrieval:

• Does the candidate address the challenge of providing contextual assistance and retrieving relevant knowledge articles or resolution steps to help resolve user queries?

• Are they able to integrate the chatbot with Jira Service Management’s knowledge base and support documentation for seamless information retrieval?

5. Intelligent Escalation and Collaboration:

• How does the candidate plan to handle complex queries or issues that require human intervention or escalation to higher-tier support agents?

• Can they suggest mechanisms for facilitating collaboration between the chatbot and human agents within Jira Service Management’s workflow?

6. Performance Monitoring and Improvement:

• What metrics does the candidate propose for evaluating the performance and effectiveness of the chatbot in improving customer support outcomes?

• Are they able to incorporate mechanisms for collecting user feedback, monitoring chatbot interactions, and iteratively refining the NLP models based on real-world usage data?

Rating Criteria:

• NLP and ML Expertise (5/5): Demonstrates proficiency in NLP and ML techniques for natural language understanding and dialogue management.

• Automation and Efficiency (4/5): Provides innovative approaches for automating routine tasks, reducing response times, and improving overall efficiency in customer support.

• Integration and Interoperability (4/5): Addresses technical challenges in integrating the chatbot with Jira Service Management’s APIs and workflows for seamless interaction and collaboration.

• User-Centric Design (4/5): Considers usability, context sensitivity, and personalized assistance to enhance user experience and satisfaction with customer support interactions.

• Impact Assessment (4/5): Identifies potential benefits of the chatbot in terms of reduced ticket resolution times, improved first-contact resolution rates, and enhanced customer satisfaction scores within Jira Service Management.

Design a Recommendation Engine for improving task management in Trello

Problem Statement:

Trello, Atlassian’s visual collaboration tool, is widely used for managing tasks, projects, and workflows. However, users often struggle to prioritize tasks and allocate resources effectively within their Trello boards. Design a recommendation engine that leverages machine learning algorithms to analyze user behavior, task attributes, and board dynamics, in order to provide intelligent recommendations for task prioritization, assignment, and scheduling within Trello.

Key Insights and Signals:

1. Understanding of Trello’s Usage Patterns:

• Does the candidate have a comprehensive understanding of how Trello boards are structured and used for task management?

• Can they identify common challenges faced by users in organizing, prioritizing, and tracking tasks within Trello?

2. User Behavior Analysis:

• How does the candidate propose to capture and analyze user interactions, task updates, and board activities within Trello?

• Are they familiar with techniques such as user clustering, behavioral segmentation, and sequence modeling for extracting meaningful insights from user data?

3. Task Attributes and Contextual Information:

• What features does the candidate suggest extracting from task cards, including due dates, labels, descriptions, and attachments, to characterize task attributes?

• Can they incorporate contextual information such as board categories, team roles, and project deadlines to improve recommendation relevance?

4. Recommendation Algorithms:

• Does the candidate propose algorithms for generating personalized recommendations for task prioritization, assignment, and scheduling based on user preferences and board context?

• Are they able to balance between simple heuristic-based approaches and more sophisticated machine learning models to ensure practical feasibility and effectiveness?

5. Integration with Trello Platform:

• How does the candidate plan to integrate the recommendation engine seamlessly into the Trello user interface and workflow?

• Can they propose API endpoints, webhooks, or browser extensions for delivering recommendations directly within Trello boards?

6. Evaluation and Feedback Loop:

• What metrics does the candidate suggest for evaluating the quality and impact of the recommendation engine on user productivity and task completion rates?

• Are they able to incorporate mechanisms for collecting user feedback and iteratively refining the recommendation algorithms based on user preferences and performance metrics?

Rating Criteria:

• Trello Expertise (5/5): Demonstrates deep knowledge of Trello’s features, usage patterns, and user needs in task management.

• Recommendation Algorithm Design (4/5): Provides innovative approaches for generating personalized recommendations tailored to individual users and board contexts.

• Practical Feasibility (4/5): Addresses technical challenges in implementing the recommendation engine within the Trello platform while ensuring scalability and performance.

• User-Centric Design (4/5): Considers usability, relevance, and integration with existing Trello features to enhance user experience and task productivity.

• Impact Assessment (4/5): Identifies potential benefits of the recommendation engine in terms of improved task prioritization, resource allocation, and team collaboration within Trello.

Products and ML Problems

1. Confluence:

Contextual Search Enhancement System for Confluence Cloud:

• How would you enhance the existing search functionality in Confluence using contextual information to improve search results?

• What techniques or algorithms would you employ to understand the context of user queries and documents?

Intelligent Q&A System for improving knowledge sharing in Confluence Cloud:

• Discuss your approach to designing a system that intelligently retrieves answers to user questions from the vast repository of knowledge stored in Confluence.

• How would you incorporate natural language understanding and reasoning capabilities to ensure accurate and relevant responses to diverse user queries?

• What strategies would you employ to handle ambiguity, synonymy, and variability in user questions and document content effectively?

Collaborative Filtering Recommendation System for Confluence Cloud:

• Explain how you would implement a collaborative filtering recommendation system to suggest relevant content to users in Confluence.

• How would you address challenges such as sparsity of user interactions and cold start problems?

Adaptive Document Summarization System for Confluence Cloud:

• How would you approach building a system that generates concise summaries of lengthy documents stored in Confluence?

• What strategies would you employ to ensure the summaries capture the essential information while maintaining coherence and relevance?

Dynamic Content Tagging System for Confluence Cloud:

• Discuss your approach to developing a system that automatically tags content in Confluence based on its context and relevance.

• How would you handle the challenge of dynamically updating tags as the content evolves over time?

Multi-modal Content Understanding System for Confluence Cloud:

• How would you integrate text, images, and other modalities of content to enhance understanding and retrieval in Confluence?

• What techniques or architectures would you consider for handling multi-modal data effectively?

Continuous Learning System for Confluence Cloud:

• Describe how you would build a system that continuously learns from user interactions and feedback to improve its recommendations and search results in Confluence.

• What mechanisms would you employ to ensure the system remains up-to-date and adaptable to changing user preferences and content dynamics?

Explainable AI Framework for Content Recommendations in Confluence Cloud:

• Discuss the importance of explainability in AI-driven content recommendation systems for enterprise applications like Confluence.

• How would you design a framework that provides transparent explanations for the recommendations made to users?

2. Jira Software:

• Issue Prioritization: ML can be used to analyze historical data on issue resolution times, dependencies, and user feedback to prioritize tasks and allocate resources more effectively.

• Sprint Planning: ML algorithms can assist in predicting the completion time for tasks and recommending optimal task assignments for sprint planning sessions.

• Automated Ticket Categorization: ML algorithms can classify incoming support tickets based on their content, urgency, and potential impact, enabling faster ticket routing and resolution.

• Customer Sentiment Analysis: ML-powered sentiment analysis can analyze customer interactions and feedback within tickets to detect sentiment trends and identify areas for improvement in service quality.

3. Bitbucket:

• Code Review Assistance: ML techniques can analyze code changes, comments, and historical code review outcomes to provide real-time suggestions and feedback during code review sessions, improving code quality and developer productivity.

• Branch Management: ML can analyze historical branching patterns, merge conflicts, and code dependencies to recommend optimal branching strategies and workflows for managing code repositories in Bitbucket.

• Code Quality Analysis: ML algorithms can analyze code repositories to identify code smells, security vulnerabilities, and best practice violations, providing actionable insights for improving code quality and maintainability.

• Codebase Health Monitoring: ML-powered bots can continuously monitor code repositories for changes in code complexity, dependency risks, and technical debt, alerting developers to potential issues and recommending corrective actions to maintain codebase health.

• Code Review Automation: ML-powered code review tools can automatically identify code quality issues, suggest code improvements, and enforce coding standards during the review process, reducing manual effort and ensuring consistent code quality.

• Continuous Integration Optimization: ML algorithms can analyze historical build and test data to optimize the configuration of continuous integration pipelines, improving build performance and reducing build failures.

4. Trello:

• Task Recommendation: ML algorithms can analyze user behavior, task attributes, and project dynamics to recommend task prioritization, assignment, and scheduling strategies within Trello boards, improving team productivity and project outcomes.

• Workflow Automation: ML-powered bots can automate routine tasks and workflows within Trello boards, such as task assignment based on workload, deadline reminders, and progress tracking.

• Workflow Optimization: ML algorithms can analyze user workflows, task dependencies, and completion times to identify bottlenecks and inefficiencies in project management processes, recommending workflow optimizations for improved team productivity.

• Predictive Task Completion: ML techniques can analyze task attributes, team dynamics, and historical completion times to predict the likelihood of task completion within specified deadlines, enabling better resource allocation and project planning.

• Project Timeline Prediction: ML algorithms can analyze historical project data, including task completion times, dependencies, and resource allocation, to predict project timelines and milestones, aiding in project planning and resource management.

• Task Clustering and Organization: ML techniques can automatically cluster similar tasks or cards within Trello boards based on their content, attributes, and relationships, helping users organize and prioritize their work more efficiently.

5. Opsgenie:

• Alert Triage: ML can help classify and prioritize incoming alerts based on severity, impact, and historical incident data, enabling faster incident response and resolution times.

• Incident Prediction: ML algorithms can analyze patterns in infrastructure metrics, user activity, and system logs to predict potential incidents before they occur, allowing proactive mitigation and preventive measures.

• Predictive Incident Resolution: ML algorithms can analyze historical incident data, including resolution times, root causes, and response actions, to predict the most effective resolution strategies for future incidents, reducing downtime and minimizing impact on operations.

• Resource Optimization: ML techniques can analyze team availability, skill sets, and workload data to optimize on-call schedules, ensuring the right resources are available to respond to incidents promptly and efficiently.

• Anomaly Detection in Monitoring Data: ML techniques can analyze real-time monitoring data from infrastructure and applications to detect anomalies, performance degradation, or security threats, triggering automated incident response actions in Opsgenie.

• Service Dependency Mapping: ML algorithms can analyze service interdependencies and communication patterns to create dynamic service dependency maps in Opsgenie, aiding in incident management and root cause analysis.

6. Statuspage:

• Service Health Monitoring: ML techniques can analyze historical data on service uptime, incident resolution times, and user feedback to predict service health and performance trends, enabling proactive communication and issue resolution.

• Incident Communication: ML-powered chatbots can assist in automating incident communication and status updates on Statuspage, ensuring timely and accurate information dissemination to stakeholders during service disruptions.

• Performance Trend Analysis: ML algorithms can analyze historical performance data, including response times, uptime, and error rates, to identify performance trends and predict potential issues before they impact service availability or user experience.

• Subscriber Communication Optimization: ML techniques can analyze subscriber engagement metrics and communication preferences to optimize communication strategies during service disruptions, ensuring timely and relevant updates to subscribers while minimizing notification fatigue.

7. Stride/Slack (formerly Atlassian Stride):

• Sentiment Analysis: ML algorithms can analyze team communication data to detect sentiment trends, identify potential conflicts or morale issues, and provide insights for improving team dynamics and collaboration.

• Automated Meeting Summarization: ML-powered bots can summarize meeting transcripts, extract action items, and highlight key discussion points, making it easier for team members to follow up on meeting outcomes and decisions.

• Automated Workflow Assistance: ML-powered bots can analyze team communication patterns and workflows within Stride/Slack channels to provide automated assistance, reminders, and notifications for upcoming tasks, meetings, or deadlines.

• Employee Onboarding Support: ML algorithms can analyze onboarding-related conversations and documentation within Stride/Slack channels to provide personalized onboarding assistance, resources, and guidance for new employees.

Sample Questions

Scrapped from the Internet

Machine Learning Concepts:

1. How would you distinguish an RNN from an LSTM in terms of structure and function?

2. Would you mind explaining the Random Forest model and its significance in predictive analytics?

3. How do you handle skewed data when evaluating model performance, and what are some common metrics used in such cases?

Dimensionality Reduction:

4. Could you describe some methods for reducing dimensionality and how they’re used in Machine Learning?

Predictive Modeling and Data Analysis:

5. Considering your experience with data analysis and client buying behaviors, can you walk me through how you would make predictions around whether future clients would purchase a certain software? What pieces of information would be important to include in your dataset?

System Architecture and Design:

6. Describe the components and design principles you would incorporate into the Bing image search architecture.

Programming and Algorithmic Skills:

7. Can you construct a function that generates a random normal distribution and then plot it?

8. Consider an array of sorted integers from 0 to n. Your task is to find the integer that introduces a problem. Write a function that accomplishes this task with a time complexity of O(log n).

9. Can you demonstrate how to reverse a binary tree in a selected programming language?

10. Given an array of integers and a target sum, find the smallest subarray with a sum greater than or equal to the target sum using the greedy approach.

11. What method would you use to find the dot product between two sparse matrices?

Behavioral and Team Dynamics:

12. What is the reason behind your search for a new job?

13. What would be your ideal team to join in Atlassian?

14. Have you ever had to object to a team member’s approach? Can you walk me through it?

15. Can you tell us about a time where you received unpleasant feedback?

16. In what ways are your experiences aligned with Atlassian’s values?

17. What are your long-term career goals, and how do you see yourself achieving them over the next five years?

18. So far, what has been your biggest accomplishment?

19. Describe a time that you took a risk?

20. Tell me about a time when you assisted a colleague in his work. What was the result?