AI has been subject to debate by plenty of large technology companies in recent months. Conversations about regulation and worldwide anxieties about the fast-paced evolution of the technology have prompted companies like Meta to commit to greater research and open source models to ensure responsible and ethical systems.
We speak with Amey Porobo Dharwadker, Engineering Leader in Machine Learning at Meta, about leveraging AI to enhance user engagement at the company. Having worked at the company since 2015, Dharwadker also offers insight into how he sees AI shaping the digital landscape moving forward into 2024.
Tell us a bit about your career background. How did you come to be working at Meta?
I earned my undergraduate degree in Electronics and Communication Engineering from National Institute of Technology Tiruchirappalli, one of India's premier technical institutions. My interest in machine learning and computer vision was sparked here at school, through various elective courses.
Following a two-year stint at Analog Devices in India, where I worked on advanced driver assistance systems using computer vision, I pursued a Master's degree at Columbia University, USA. I had the privilege of interning at Meta (formerly known as Facebook) at their headquarters in Menlo Park, California. As an intern, I primarily focused on Ads Ranking, optimising ads relevance to maximise value for both - advertisers and users.
This experience laid the foundation for my journey into large-scale machine learning challenges. The vibrant work culture, autonomy and the opportunity to collaborate with some of the smartest people in the industry drew me in. Consequently, I decided to join Meta full-time in January 2015 upon my graduation to work on building personalisation models for News Feed recommendations.
Can you elaborate on your role in leveraging AI at Meta to enhance user engagement on a global scale?
My role at Meta is centred around leading the Facebook video recommendations ranking team. I'm deeply involved in the continuous improvement of Facebook's Videos and Reels recommendation systems, impacting the daily experiences of over two billion users globally.
A notable highlight of my journey is that I was one of the first machine learning engineers involved in launching the Video tab on Facebook back in 2016.
Over my nine year tenure at Meta, I have enjoyed working across diverse domains; from shaping personalisation models for News Feed recommendations to steering initiatives that enhance ads attribution models. I have introduced significant improvements to our Facebook video recommendations ranking personalisation models since 2019.
Leveraging neural modelling and multi-task learning among other techniques, these advancements have transformed the way billions of users discover and engage with relevant video content on the platform.
Our team has recently published research addressing conformity bias and exploring user interests in billion-users scale industrial recommender systems. This work holds broad implications for advancing fairness in the multi-stakeholder recommendation marketplace.
At its core, I find immense fulfilment in working within the realm of technology, firmly believing that significant advancements in humanity stem from transformative technological shifts. This conviction is the driving force behind my commitment to success in this field, as I perceive technology as the primary catalyst for positive global change, and I'm enthusiastic about being an integral part of that transformative journey.
How do you see AI shaping the future digital landscape?
From my vantage point in the world of AI-driven recommendation systems, I foresee an exciting future for personalisation. We're moving beyond just understanding user preferences, behaviours and contexts - the next frontier is even more finely tuned and context-aware personalisation, offering users tailored and immersive digital journeys.
The democratisation of AI is another exciting trend. As technology advances, we're witnessing a broader accessibility to advanced AI tools and frameworks. This has the potential to empower smaller businesses and individual developers to innovate and shape the digital landscape.
It's not just about what AI can do for large tech companies, it's about how it can catalyse innovation across the digital spectrum. I'm betting on an automation acceleration in the coming years, freeing up human potential for more strategic and creative endeavours. That's the AI-driven future I'm stoked about.
In what ways do you contribute to the field outside of your primary role?
Beyond my primary role, I thrive on being deeply woven into the fabric of the AI and tech community. I give talks at top-tier conferences sharing insights on cutting-edge AI technology and advise and mentor startups. I'm also a regular program committee member and reviewer for esteemed international conferences and journals in machine learning, actively helping shape the field's trajectory.
Co-organizing the VideoRecsys workshop at the ACM Conference on Recommender Systems, I have contributed to creating a platform for researchers, practitioners and industry experts to delve into the latest trends in large-scale recommendations. This collaborative platform not only fuels the recommender systems community but also sparks future research ideas in this dynamic field.
Drawing from your career experiences, what advice would you give those starting in the field of AI?
When it comes to thriving in the fast-paced world of AI and machine learning, I would emphasise the importance of building a rock-solid foundation in the fundamentals. The allure of cutting-edge models and complex architectures is hard to resist, but it's the basics that truly empower innovation.
I recall my early days, spending hours delving into the core principles of machine learning and wrapping my head around the underlying mathematical concepts. It was the bedrock for everything that followed. Don't rush through the fundamentals. Treat them like the essential toolkit they are, they pave the way for diving into more advanced concepts and methodologies with confidence.
Secondly, immerse yourself in real-world applications. Theoretical knowledge is essential, but the true power of AI is unleashed when you leverage it to address practical challenges. I encourage you to seek out projects that align with your interests and actively contribute to open-source communities. This hands-on experience will not only deepen your understanding but also put a spotlight on your skills for potential employers.
What else do you hope to learn in your career and how do you envision expanding your skills?
I'm keen on immersing myself in the latest indigenous tools and technologies within the ever-evolving AI and machine learning landscape.
My pursuit extends beyond mere intellectual curiosity, it is a pragmatic necessity. My goal is to seamlessly integrate these technologies into our production systems, ensuring they're not just state-of-the-art but also scalable to billions of users globally, providing a truly personalised and relevant digital experience.
I also have a deep interest in gaining a nuanced understanding of the ethical considerations surrounding AI. As the influence of this technology continues to grow, ensuring fairness, transparency and accountability in our algorithms is paramount.
I envision actively participating in shaping industry-wide ethical guidelines, fostering a culture of responsible AI development and contributing to the ongoing discourse surrounding the ethical implications of delivering personalised content at scale.
AI Magazine is a BizClik brand