Thank you so much for joining us in this interview series. Before we dive into our discussion our readers would love to “get to know you” a bit better. Can you share with us the backstory about what brought you to your specific career path?
Ofcourse! My journey into the field of AI and machine learning began with a deep fascination for language and technology. Growing up, I was always intrigued by the way language shapes our understanding of the world. This curiosity led me to pursue a Master’s in Theoretical Computer Science at Stanford University, where I studied the fundamentals of natural language processing (NLP) and artificial intelligence (AI).
After graduating, I joined Google, where I worked on the Speech team to deploy automatic speech recognition (ASR) models at scale. This experience was incredibly rewarding, as it allowed me to see firsthand AI’s impact on improving communication and accessibility for billions of users worldwide.
My time at Google also reinforced my belief in the power of multilingual communication, which eventually led me to Welocalize. As the Head of AI/ML Engineering, I am dedicated to developing innovative solutions that enhance translation and localization processes, making it easier for people from diverse linguistic backgrounds to connect and collaborate.
Throughout my career, I’ve been driven by the desire to bridge language barriers and create technologies that foster global understanding. At Welocalize, I continue to pursue this mission by leveraging cutting-edge machine learning technologies to ensure seamless and accurate translations for global audiences.
What do you think makes your company stand out? Can you share a story?
From a technical perspective, it might at first glance seem like the toolsets we offer are similar to competitors. However, the devil is in the details when it comes to performance. Welocalize has always had rigorous standards for quality, and we focused on building AI tools into our Opal product suite that can perform at human-level. We have put in an enormous amount of engineering work over the past 6 months with our brightest minds to ensure that both our AI Quality Estimation (AIQE) and AI Post Editing (AIPE) are best in class. For example, differentiator for our AIPE model, versus our competitors, is that we built a very sophisticated system for retrieval (RAG) which allows the AI Post-Editing (AIPE) to mimic how a human would work, using all the information that the human post editor would have at hand and retrieving relevant information from various client resources.
In comparison, we suspect that others in the language services and technology industry have only a generic AIPE with prompts like “translate to German in a positive tone”. This is nice, but serves as more human-augmentation, rather than a human-replacement.
You are a successful business leader. Which three character traits do you think were most instrumental to your success? Can you please share a story or example for each?
I prefer to attribute success to habits and behaviors more than character traits, since habits are buildable. There are three key habits I would highlight: scientific discipline and persistence, focusing on the vision, and building a team better than myself. These have consistently guided my decisions and actions, enabling me to overcome challenges and drive meaningful progress.
- Scientific Discipline and Persistence
When we were developing our AI-LQA model, we faced significant hurdles. Despite extensive training data and prompt tuning, achieving high accuracy relative to human data was elusive. Under pressure to deliver, I had to admit to the C-suite that after significant research, we still hadn’t succeeded. This honesty led us to broaden our labeling and data analysis efforts, ultimately revealing a crucial insight: human LQA evaluators often disagree, indicating the inherent difficulty of the task. This breakthrough, now validated by external researchers, was only possible because we remained disciplined and persistent, pushing through until we had a definitive answer.
2. Focusing on the Vision
Maintaining a clear vision is essential. It’s easy to get caught up in immediate tasks, but regularly aligning with our strategic and technical vision ensures that we focus on what truly drives progress. By consistently referring to our 1-year, 5-year, and 10-year plan, we prioritize effectively, focusing on initiatives that move us forward and avoiding distractions that don’t contribute to our long-term goals.
3. Building a Team Better Than Yourself
Our success is a testament to the strength of my team. They are the true architects of our achievements, while I provide strategic direction. A leader’s effectiveness is amplified by a capable and reliable team. I am fortunate to work with individuals who excel in their roles, and together, we achieve far more than any one person could alone. Building a team that is better than yourself is fundamental to driving innovation and success.
Let’s now move to the main point of our discussion about AI. Can you explain how AI is disrupting your industry? Is this disruption hurting or helping your bottom line?
We have a thriving business line, Welo Data, which focuses on the creation of high-quality datasets for training AI models. We are busier than ever!
The localization work brings both challenges and opportunities. In the past year or so, these tools have advanced to the point where, with a fair bit of clever engineering, they can often perform at the level of a human. We see that with our AI-Post-Editing (AIPE) tool.
That has profound implications for the industry. The place of the human in the workflow is radically shifting. We do still believe that there is a need for humans in localization workflows, but if you add finely calibrated AI like Welocalize has built, you very rarely need two humans to look at every word you translate. You can use AI systems like our AIPE to ensure that the translation aligns with client-specific standards, and then you can use AIQE to direct extra human attention at only the most nuanced or highest-risk segments.
Which specific AI technology has had the most significant impact on your industry?
I know it’s old news now, but machine translation represented a huge shift in the industry. The idea that we could use machines to do the bulk of the translation work, and just use humans to correct and refine the output, was a fundamental shift in the way we think about localization at that time.
That trend of allowing AI/ML models to do more of the work has just accelerated with the release of highly performant LLMs like GPT, Gemini, Claude and similar. With these models as a basis, and AI/ML engineering work to specialize them to specific localization tasks, we can make huge strides in efficiency in the day-to-day localization work.
Can you share a pivotal moment when you recognized the profound impact AI would have on your sector?
As someone relatively new to the language services industry, I’ve only known it as a dynamic and rapidly evolving field. The industry has been undergoing significant transformations for years, driven by continuous advancements in AI technologies.
In recent times, Large Language Models (LLMs) and Generative AI (Gen AI) have emerged as game-changers. These technologies have revolutionized the way we approach translation and localization, enabling unprecedented levels of speed, accuracy, and efficiency. They go beyond simple word-for-word translation, understanding context, cultural nuances, and maintaining the tone of the original content. This evolution solidifies AI’s role as a disruptive force in the language services sector, driving innovation and setting new standards for the industry.
How are you preparing your workforce for the integration of AI, and what skills do you believe will be most valuable in an AI-enhanced future?
My team is mostly AI/ML engineers, so they are quite prepared. However, for the less technical folks, I think that writing and clear communication have become even more paramount when prompting an LLM. Getting an LLM to augment your productivity requires explaining your goals in a clear and thorough way — honestly, that’s the core of most prompt engineering.
What are the biggest challenges in upskilling your workforce for an AI-centric future?
The biggest challenge is making time for upskilling in a structured and coordinated way. People usually want to learn new skills and be more efficient, but it’s important to lead the organization in a way that gives them space, time, and incentive to learn the tooling.
What ethical considerations does AI introduce into your industry, and how are you tackling these concerns?
For all AI and ML models, I think there is a persistent and real concern around data privacy and use constraints. Here at Welocalize, we consider data privacy to be of utmost importance. We never send any data to external models unless we have robust enterprise agreements in place which ensure that data cannot be used or stored by those providers. And we have very high data engineering and privacy standards to ensure that our clients’ data is only being used in legal, safe, and appropriate ways.
What are your “Five Things You Need To Do, If AI Is Disrupting Your Industry”?
- Focus on Data Quality and Integrity: Ensuring high-quality and accurate data is critical for effective AI implementation. Data is “fuel” for AI models. Invest in data engineering across the organization, to ensure that critical data is collected, stored, well-organized, and usable. At Welocalize, we were able to create an entirely new, award-winning AI-enabled product suite (Opal) in 6 months, and there is no way we could have done that without extensive data engineering.
- Invest in AI Expertise. Ensure that you have people on your team that are experts in, and champions for, AI. Keep them, reward them for their efforts, and elevate their voices.
- Recruit Experts: Hire data scientists, machine learning engineers, and AI researchers. Look for professionals with experience in your industry’s specific challenges.
- Train Your Team: Upskill your current workforce with AI and machine learning training programs to ensure they can work effectively with new technologies. For example, Welocalize has done extensive internal training to ensure that all folks at the company are empowered to use AI in a safe and effective way.
3. Understand the AI Landscape
- Stay Informed: Keep up with the latest AI research, tools, and trends specific to your industry. Follow key publications, attend conferences, and participate in webinars. For example, I carve out time to look at new research papers every week and try to get through the 5–10 that look most relevant for our work. This ensures that we can thoughtfully and rapidly test new inventions and incorporate the latest tech into our workflows.
- Competitive Analysis: Analyze how competitors are leveraging AI and identify best practices and emerging opportunities.
4. Adapt and Scale: Plan for scalability from the outset. Design AI solutions that can grow with your business and handle increasing data volumes and complexity.
Monitor and Measure Impact
- KPIs and Metrics: Choose your KPIs carefully, to be able to understand and measure what kind of impact you are looking to glean from any AI initiatives.
- Continuous Improvement: Adopt a continuous improvement mindset, using data-driven insights to refine and optimize AI models and processes.
What are the most common misconceptions about AI within your industry, and how do you address them?
One of the most common misconceptions about AI in the language services industry is that all human linguists’ jobs will evaporate. While it’s true that AI is transforming the way we work, it is not eliminating the need for human expertise.
AI, particularly in localization and translation, relies heavily on human expertise to train and refine the models. Experts in the loop are essential for several reasons:
- Ongoing creation of training data
- Quality Assurance
- Handling Complex and Nuanced Tasks
By integrating AI into current workflows, we are not eliminating jobs but rather transforming them. The role of human experts is evolving to include overseeing AI processes, improving AI systems, and handling tasks that require human intuition and creativity. This collaboration between humans and AI allows us to achieve higher efficiency and quality in our work, ultimately benefiting both our clients and our employees.
Can you please give us your favorite “Life Lesson Quote”? Do you have a story about how that was relevant in your life?
Honestly, I’m not one to preach. I guess my best advice is to wear sunscreen and get good sleep.
Off-topic, but I’m curious. As someone steering the ship, what thoughts or concerns often keep you awake at night? How do those thoughts influence your daily decision-making process?
I would say if I’m being kept awake, it’s usually excitement about the amazing things Welocalize is building. In general, as a technical leader, I would generally worry about robustness, security, data privacy, downtime, things like that. But when I joined Welocalize we already had an incredibly strong and mature platform, high engineering standards, and excellent data practices, so I can sleep soundly.
You are a person of great influence. If you could start a movement that would bring the most amount of good to the most amount of people, what would that be? You never know what your idea can trigger. 🙂
I would focus on addressing climate change through the lens of social and economic inequality. I’d prioritize marginalized communities, ensuring they have access to clean energy, resilient infrastructure, and sustainable livelihoods. By investing in renewable energy, promoting sustainable agriculture, and empowering vulnerable populations, we can mitigate climate change and reduce systemic inequalities.
How can our readers further follow you online?
https://www.linkedin.com/in/mikaelagrace/
Thank you for the time you spent sharing these fantastic insights. We wish you only continued success in your great work!