EN

#

Technical Interview Guide

This guide was created to help you prepare for QuintoAndar's technical interviews, with detailed tips, study suggestions, examples, and free resources.

# Data Scientist

Here you find the tips and guidelines we have prepared for the technical interviews in the Data Scientist area.

Choose where to start or review everything at your own pace. Happy reading and happy preparing!

ML and AI Modeling

This is our first technical conversation! We will discuss your experience and core concepts of Data Science, Machine Learning, and AI. The goal is to understand your reasoning and vision. If we have a great session here, you move on to the next steps!

      📚 How to prepare – ML and AI modeling:

      Review previous projects: Summarize your most important experiences in ML/DS/AI.

      Technical Fundamentals: Be ready for questions about algorithms and concepts you’ve used. Explain the intuition behind your techniques.

      Example: Have you worked with Gradient Boosting? If so, can you intuitively explain how boosting models work and how they differ from bagging?

      Business Case Approach: You will solve a real-world case focused on problem structuring. Use an iterative approach: start with an MVP (end-to-end) and gradually add complexity and corner-case handling. We value logical clarity over just finding the “right” answer.

      📌 Checklist – ML and AI modeling:

      Want to make sure you haven’t forgotten anything? Use this checklist to guide your approach:

      [ ] I understand the core ML methodologies and concepts, as well as the techniques and methods applied in my projects. I am prepared for technical questions regarding ML, DS, and AI.

      [ ] I have relevant Data Science, AI, or Machine Learning project(s) ready for discussion, and I can clearly explain my past experiences in these disciplines.

      [ ] I can explain the intuition behind the techniques I applied in my projects.

      [ ] I can demonstrate my reasoning skills within a business case.

      [ ] I know how to structure a problem and think through different solutions.

      📑 Support Materials – ML and AI modeling:

      • The Hundred-Page Language Models Book – Andriy Burkov
      • The Hundred-Page ML Book  – Andriy Burkov
      • Intro to ML – Kaggle 
      • Intermediate ML – Kaggie
      • AI Engineering: Building Applications with Foundation Models – Chip Huyen
      • Build a Large Language Model (From Scratch) – Sebastian Raschka
      • Transformers: The Definitive Guide – Nicole Koenigstein

DS Case

In this stage, we model a problem based on real QuintoAndar scenarios and dive deep into ML topics.

  • 💙 What we expect:

    • Technical mastery (Classification, Regression, Metrics, Monitoring, Feature Engineering, Agentic AI evals, Prompt Engineering, Deep Learning).
    • Strategic reasoning and business-oriented results.
    • Good storytelling.

      📚 How to prepare – DS Case

      1. Technical Domain and Fundamentals

      What we expect: We value those who understand the mechanics behind the tools. When proposing a solution, do not limit yourself to simply citing a model or algorithm; demonstrate depth by detailing the inner workings of the chosen methods or metrics.

      • Example: Instead of just naming an algorithm (“I will use Model X”), explain why its loss function is suitable for your data distribution or how the chosen evaluation metric directly connects to business success.

      2. Focus on the Solution (Agnostic Approach)

      What we expect: The primary goal is to solve the problem effectively and sustainably. The method is a tool to achieve the result; we want to see your ability to choose the approach that best balances performance and viability.

      • Example: When facing a sorting or recommendation problem, the focus should be on how you build a robust solution. More important than using the most complex architecture is ensuring that the strategy solves the user’s pain point and that you know how to measure that impact.

      3. Communication and Collaborative Reasoning

      What we expect: We want to know your thought process in real-time. View the interviewer as a project partner, promoting a constant exchange of ideas and validation of assumptions.

      • Example: During resolution, share your trade-offs transparently: “I am opting for this initial structure to validate the main hypothesis, but in a scaling scenario, we could follow path Y. Does this prioritization make sense to you?”.

      4. Incremental Delivery (Solve, then Optimize)

      What we expect: Demonstrate pragmatism. Start by establishing a solid, functional solution that addresses the core of the problem (“end-to-end”) before moving on to discuss and implement technical refinements.

      • Example: If the challenge involves modeling, first structure a complete pipeline with a reliable base model. Once the logical flow is validated, demonstrate how you would evolve to more sophisticated techniques (such as contrastive learning or specific architectures) to gain extra performance.

      📌 Topics for Technical Interview – DS Case

      • Modeling problem definition, including objective formulation and its technical and business impacts.
      • Machine Learning model development, covering modeling and training decisions.
      • Data use and preparation, focusing on variables and information used by the model.
      • Model evaluation, considering technical criteria and experimentation.
      • Interpretation of results, understanding model behavior.
      • Model application in a production environment, including system integration and result monitoring.

Coding and SQL

In this interview, you will need to use algorithms and SQL to solve one or more programming problems in a Machine Learning context. This is the moment to show your coding skills in practice!

  • 📑Supporting Materials

    • Super Study Guide: Algorithms & Data Structures – Afshine Amidi & Shervine Amidi

      📚 How to prepare – Coding & SQL

      The goal is to evaluate your ability to apply logic and programming to build the foundation of a data project.

      Programming Languages

      • Feel comfortable with at least one programming language of your choice. It is important to know Object-Oriented Design and how to test your code. For data science and ML at QuintoAndar, we use Python.
      • You should know how to create methods and functions to apply ML models to policies or format the outputs of ML models or LLM calls to solve a business problem.

      Data Structures

      • Familiarize yourself with arrays (lists), linked lists, stacks, queues, sets, maps (maps/dictionaries), trees, and graphs.

      Complexity

      • Demonstrate that you can reason about the time and space complexity of an algorithm and how to improve it. It is good to know the basics of Big-O notation.

      Dataset Creation

      • Be able to query data using best practices to build a training dataset. Be able to set up transformations to create features or data aggregations.

      ✅ Good Practices and Correctness:

      • You will use an online IDE with basic features. Take advantage of code completion and syntax highlighting to test your code.