1. What is a Data Scientist at Tessian?

At Tessian, a Data Scientist plays a pivotal role in the company’s mission to secure the "Human Layer." unlike traditional cybersecurity approaches that focus on networks or devices, Tessian uses machine learning to protect people from advanced email threats like phishing, spear-phishing, and accidental data loss. This role sits at the intersection of cybersecurity, natural language processing (NLP), and behavioral anomaly detection.

You will not just be analyzing static datasets; you will be building and deploying models that must make split-second decisions on live email traffic. The work involves tackling complex problems such as detecting brand impersonation, understanding relationship graphs between employees, and identifying subtle textual cues that indicate a security threat. The impact of your work is immediate—a single model improvement can stop thousands of malicious emails from reaching inboxes.

This position requires a blend of research capability and engineering mindset. You will work closely with engineering teams to productionize your models, ensuring they are scalable and low-latency. For candidates, this represents a challenging but rewarding opportunity to apply advanced ML techniques to high-stakes, real-world problems where accuracy and speed are paramount.

2. Getting Ready for Your Interviews

Preparing for the Tessian interview process requires a shift in mindset from standard data science interviews. You need to be ready to demonstrate not just theoretical knowledge, but also the ability to write production-ready code and architect systems from scratch.

Technical Proficiency – You will be evaluated heavily on your ability to write clean, efficient Python code. Unlike some data science roles where scripting is enough, Tessian expects strong algorithmic fundamentals. You must be comfortable implementing logic that goes beyond library imports, such as parsing expressions or manipulating complex data structures.

Problem Structuring & Product Sense – Tessian places a high value on how you approach ambiguous problems. You will likely face case studies related to email security (e.g., "How would you detect if an email is fake?"). Interviewers are looking for a structured approach: identifying data sources, recognizing data quality issues (dirty data), designing the high-level flow, and iterating on edge cases.

ML System Design – For mid-to-senior roles, you must demonstrate an understanding of the end-to-end ML lifecycle. This includes discussing architecture, how to serve models in real-time, and how to handle feedback loops. You should be prepared to sketch out how a solution fits into a larger platform, rather than just describing a specific algorithm.

Communication & Collaboration – You will often be interviewed by the people you will work with daily. They assess whether you can explain complex technical nuances to cross-functional stakeholders. Being able to articulate why a specific model is the right choice—and acknowledging its limitations—is just as important as the choice itself.

3. Interview Process Overview

The interview process at Tessian is generally regarded as rigorous and structured, with a difficulty level often described as "Hard" or "Very Hard." While historical data suggests some variability in the candidate experience regarding feedback timelines, recent reports indicate a well-organized process designed to test relevant skills deeply. You should expect a multi-stage funnel that moves from automated screening to intense interactive problem-solving.

Typically, the process begins with a recruiter screen followed by a technical assessment. This assessment is often a HackerRank-style coding challenge focused on Python and algorithms. If you pass this stage, you will move to a series of virtual onsite rounds. These rounds are distinct and cover specific competencies: a deep-dive technical interview (often involving a case study), a system design or architecture discussion, and behavioral interviews to assess culture fit and collaboration style.

The philosophy behind this process is to simulate the day-to-day work at Tessian. The coding questions are not just abstract puzzles but often relate to parsing or data manipulation tasks you might encounter in backend systems. Similarly, the case studies are drawn directly from the challenges of detecting email threats. You should be prepared for a process that tests your stamina and your ability to think on your feet under pressure.

This timeline illustrates the typical progression from application to final decision. Note that the "Technical Assessment" is a critical gatekeeper; rigorous preparation for this automated stage is essential to unlocking the face-to-face rounds. The final stage is usually a cluster of interviews, so ensure you manage your energy levels for a half-day of engagement.

4. Deep Dive into Evaluation Areas

Candidates are evaluated across three primary pillars: Algorithmic Coding, ML Case Study/Design, and Behavioral Fit. Understanding the specifics of these pillars is key to success.

Algorithmic Coding & Python

This is the most strictly evaluated technical filter. You are expected to be proficient in Python and capable of solving algorithmic problems under time constraints.

• Why it matters: Tessian’s models must run efficiently at scale. Data Scientists are expected to contribute production-quality code.
• Evaluation: Focus is on correctness, efficiency (Big O notation), and code cleanliness.
• Strong performance: You solve the problem, handle edge cases, and write readable code that follows PEP 8 standards.

Be ready to go over:

• String Manipulation & Parsing: Evaluating mathematical expressions or parsing text strings.
• Data Structures: Using stacks, queues, and dictionaries effectively.
• Time Complexity: Explaining the efficiency of your solution.

Example questions or scenarios:

• "Write a program to evaluate prefix or infix mathematical expressions." (This is a frequently reported question type).
• "Given a stream of characters, identify specific patterns or validity."

ML Case Study & Problem Solving

This is often the core of the onsite interview. You will be presented with an open-ended problem relevant to Tessian’s domain.

• Why it matters: It tests your ability to translate a vague business problem (e.g., "stop fraud") into a concrete technical solution.
• Evaluation: Interviewers look for a structured approach: Data -> Feature Engineering -> Model Selection -> Metrics -> Deployment.
• Strong performance: You proactively identify issues with "dirty data" before jumping to modeling. You can draw a high-level flow diagram and discuss trade-offs.

Be ready to go over:

• Anomaly Detection: Approaches for identifying outliers in user behavior.
• NLP Techniques: Tokenization, embeddings, and classification for text.
• Data Quality: Identifying biases or noise in a provided dataset.

Example questions or scenarios:

• "How would you build a system to detect brand impersonation in emails?"
• "Here is a sample dataset of email headers and bodies. What issues do you see with this data?"
• "Design a flow to flag emails that look like they come from a CEO but don't."

System Design & Architecture

For more senior roles, this section tests your ability to think big picture.

• Why it matters: Models do not exist in a vacuum. They must integrate with the broader Tessian platform.
• Evaluation: Ability to design scalable, reliable systems.
• Strong performance: You discuss latency, throughput, database choices, and how to monitor model performance in production.

Be ready to go over:

• ML Platform Design: How to automate training and deployment pipelines.
• Real-time vs. Batch: When to score emails (synchronously vs. asynchronously).

5. Key Responsibilities

As a Data Scientist at Tessian, your day-to-day work revolves around the intersection of data exploration and software engineering. You are responsible for the end-to-end lifecycle of machine learning models. This starts with exploratory data analysis on vast amounts of email and communication data to understand attack patterns. You will spend significant time cleaning and curating these datasets, as real-world email data is notoriously "dirty" and unstructured.

Once the data is prepared, you will design and train models—often utilizing NLP and graph theory—to detect specific threats like spear-phishing or accidental data exfiltration. However, the responsibility does not end at model training. You will collaborate closely with platform engineers to deploy these models into production environments where they must operate with low latency. You will also monitor these models to ensure they adapt to shifting attack vectors, requiring a continuous loop of feedback and improvement.

6. Role Requirements & Qualifications

To be competitive for this role, you need a strong foundation in both the theory of machine learning and the practice of software development.

• Must-have skills:

  • Python Mastery: You must be fluent in Python, including standard libraries and data structures. This is non-negotiable for passing the initial screens.
  • Applied ML Experience: Experience building models for NLP, classification, or anomaly detection problems.
  • Production Mindset: Experience or a strong willingness to write code that is deployed to production, not just Jupyter notebooks.
  • Communication: Ability to explain complex probabilistic outcomes to non-technical stakeholders.

• Nice-to-have skills:

  • Cybersecurity Domain Knowledge: Understanding of phishing, social engineering, or email protocols (SMTP, headers).
  • Graph Data Science: Experience working with network graphs (e.g., analyzing relationships between employees).
  • Cloud Platforms: Familiarity with AWS or similar cloud environments for deploying ML services.

7. Common Interview Questions

The following questions are representative of what you might face at Tessian. They are drawn from actual candidate experiences and reflect the company's focus on algorithmic soundness and domain-specific application. Do not memorize answers; use these to identify the types of thinking required.

Coding & Algorithms

• "Implement a calculator that evaluates a prefix expression (e.g., + * 2 3 4)."
• "Convert an infix expression to a postfix expression."
• "Given a list of strings, group them by a specific similarity metric."

Domain-Specific Case Studies

• "We want to build a detector for 'Brand Impersonation.' How would you approach this from scratch?"
• "Look at this sample data of email senders and recipients. What anomalies or data quality issues do you notice immediately?"
• "How would you determine if an email from 'paypaI.com' (with a capital 'i') is malicious vs. a typo?"

System Design & Behavioral

• "Design the architecture for a machine learning platform that handles millions of emails per day."
• "Tell me about a time you had to explain a model's failure to a non-technical stakeholder."
• "How do you prioritize which model improvements to work on when you have multiple options?"

8. Frequently Asked Questions

Q: How difficult is the Tessian interview process? Most candidates rate the difficulty as "Hard" or "Very Hard." The combination of a strict time-limit coding test and open-ended, deep-dive case studies makes it a rigorous process. You should not expect to "wing" the technical rounds.

Q: What is the timeline from application to offer? The process can take anywhere from 2 to 4 weeks. While some past candidates reported delays or "ghosting" in earlier years (2018-2021), more recent experiences (2022) suggest a much more organized and structured timeline.

Q: Is the coding test strictly data science focused? No. The HackerRank test often leans towards general software engineering and algorithms (e.g., parsing expressions) rather than just pandas/numpy manipulation. Treat it like a software engineer coding interview.

Q: Can I work remotely? Tessian has hubs in London and Amsterdam, and also hires for remote positions depending on the specific team and role requirements. The interview process is typically conducted entirely virtually.

9. Other General Tips

Master the "Prefix/Infix" Problem: Multiple candidates have reported facing coding questions related to parsing mathematical expressions (prefix, infix, postfix). This seems to be a standard question in their bank. Ensure you are comfortable using Stacks to solve these types of parsing problems in Python.

Focus on "Dirty Data": In the case study round, don't assume the data is perfect. Tessian deals with messy email headers and text. Explicitly look for and mention potential data quality issues (nulls, encoding errors, typos) during your analysis. This shows you have real-world experience.

Think Like an Attacker: When discussing solutions, try to adopt an adversarial mindset. If you build a model to detect X, how would a hacker bypass it? Mentioning these "adversarial attacks" and how you would mitigate them demonstrates deep domain thinking.

10. Summary & Next Steps

The Data Scientist role at Tessian is a high-impact position that demands a unique combination of algorithmic strength, machine learning expertise, and a product-focused mindset. You will be challenged to build systems that protect organizations from the most sophisticated human-centric attacks. The work is difficult but technically fulfilling, offering the chance to work with large-scale graph and text data in a production environment.

To succeed, focus your preparation on two main fronts: algorithms and system design. Polish your Python skills to the point where you can solve parsing problems comfortably under time pressure. Simultaneously, practice breaking down vague security problems into concrete data science workflows. Be ready to defend your architectural choices and discuss how your models interact with the real world.

The compensation data above provides a baseline for expectations. Note that Tessian competes for top talent in major tech hubs like London and Amsterdam, and packages often include significant equity components. Use this data to benchmark your expectations, but remember that total compensation will heavily depend on your performance in the system design and technical deep-dive rounds.

If you prepare thoroughly and approach the process with curiosity and rigor, you have a strong path forward. Good luck!