Discussion on upgrading and tuning Grammarly's ML training platform to a scalable system. Topics include moving away from a custom architecture due to hardware shortages, key requirements and architectural challenges, MLOps best practices for scalability, and lessons learned from transitioning from a single-region AWS setup to a cross-region, multi-cloud cluster compute deployment.

In this presentation, Ill tell you about my personal journey through the IT landscape, starting in Greece, moving to Canada, and then to Germany. Ill share experiences of encountering diverse leadership styles, from direct approaches to those that foster autonomy, and the importance of effective communication. During my presentation, Ill:

• Talk about techniques for adapting and the tools a leader should have to strengthen and increase team members confidence
• Spotlight the significance of inclusive tech leadership
• Highlight strategies for recognizing strengths, providing guidance, and promoting open communication with team members
• Showcase the transformative impact of embracing diversity in leadership and of fostering a collaborative and innovative tech environment

What do you do when you get a new role that you actually feel underqualified for? In 2020, I took on a new role that put me in my former companys highest-level leadership group and gave me primary responsibility for handling difficult HR-related situations. During the talk, Ill share what Ive learned:

• How I ended up in the stretch role
• How it played out and what I learned
• Leadership: key communication learnings and a mindset shift
• High-level HR responsibility: key communication learnings and an expanding repertoire
• The importance of trusted supporters and role models who model effective feedback and communication behaviors

Together, well brainstorm ways you can empower yourself and others through communication.

Navigating the challenges of communication in a fast-paced, high-energy product and tech environment when you are an introvert is not for the faint of heart. Join me to hear about my personal journey and discover how a creative, introspective mind can leverage their quiet nature as a powerful tool for effective communication and innovation, whether in the classroom or the workplace.

In this talk, Ill share:

• A definition of introversion and hypersensitivity and common challenges and misconceptions
• How expressing yourself in the community and the classroom can prepare you to communicate professionally
• Strategies for harnessing the power of introversion to improve communication skills and master the key competencies of a product manager
• Lessons on the power of quiet leadership and continuous learning on career and personal growth

This talk will help you understand the main responsibilities of analytical and computational linguists at the company, the types of tasks and projects they work on, and how they collaborate with the project teams. You will learn what kind of linguistic expertise is required for building AI-powered solutions at Grammarly.

A fireside chat exploring the full range of Grammarly's development work, from assessing potential market fit for new product offerings to deploying fine-tuned large language models into production.

Code sharing between native and web clients is never simple. Although many technologies try to solve this problem, not all are easy to use. In this talk, I will share our experience using Kotlin Multiplatform in our front-end projects using our real production feature as an example.\n\nDuring the talk, I’ll share our insights into:\n- why we chose Kotlin Multiplatform over the alternatives\n- the pain points we experienced and how we resolved them\n- what we learned about the technology’s applicability and maturity

This talk describes the newest trend in speech synthesis, where large autoregressive transformers push forward the quality of artificial speech. We will explore new emerging capabilities of state-of-the-art models and observe peculiar analogies with large language models.

Let’s delve into how Grammarly took steps to ensure user security by making FIDO2 mandatory for all team members. We will discuss the program phases and share takeaways that will help you streamline implementation. We’ll cover the following: 1) How internal trials can help you refine documentation, identify obstacles, and address areas of confusion 2) How to employ Slack bots to encourage enrollment 3) How to track and report enrollment progress 4) How to effectively tackle the implementation challenges that arise with the spread of the work-from-home work model

During this session, we’ll take a glimpse into the technical elements of planning and implementation. We’ll discuss the following topics: 1) Technical approach to selecting authentication devices based on your requirements and ideas 2) Understanding the scope of changes needed for implementation 3) Dealing with systems that are non-compatible with FIDO2 4) Organization of migration from old factors to FIDO2

Session hijacking and two-factor authentication (2FA) compromise were a painful headache for security teams for years. One-time password (OTP) authentification is common yet is insufficient to protect against modern real-world attack scenarios. According to the 2023 State of the Phish report by Proofpoint, 84% of organizations faced at least one successful phishing attack last year, while 54% faced three or more attacks. To improve the authentication process, Grammarly’s security team worked with a white-hat partner to simulate a sophisticated cyberhack on the existing OTP system. We’ll share our insights from this experiment and explain why we made the switch to the FIDO2 standard based on our learnings.

Understanding senior technical career paths can be challenging. In this talk, we will introduce the Staff Eng archetypes to help explain the career progression of individual contributors. We will also share practical guidance on navigating these paths and how they map to roles at Grammarly.

Why do we chronically underestimate the amount of time an engineering project will take? I’ll discuss some reasons why and provide strategies for making estimates that are closer to reality.

Grammarly’s team will share hands-on experience with Voice Composer, including a technical overview of Voice Composer, data, modeling, and engineering approaches, challenges encountered, and learnings for building and deploying generative AI products.

Over my four years at Grammarly, I have participated in different stages of team development, including:

• Joining a performing team as a new manager
• Building teams from scratch
• Working with teams distributed in different locations
• Switching from in-person to remote-first operating mode

Over these years, we launched three new product offerings, built many complicated features, and supported millions of users. In this talk, I want to share my experience building productive teams that achieve audacious goals.

When I joined Grammarly, it could fit in a small apartment. My first task was to fix a scalability problem: the service crashed with . . . 300 online users. Ten years later, millions of users relied on Grammarly apps everywhere—in browsers, on Windows, Mac, mobile, at home, in the enterprise, and through partnerships. Our product footprint grew, and so did our team. Along the way, we realized that scaling a software company is harder than scaling servers on AWS. As we kept hiring engineers and other essential experts, we began asking ourselves: Are we really moving faster with this big team? Are we productive? Can we even measure our developers' productivity? In this meetup I’ll tell you about engineering processes, software tools, and even the computer science principles we tried on our quest to scale engineering productivity and ease the growing pains. Join our conversation and have fun guessing: What worked and what didn't?