The ML Product Manager: Building AI-powered Solution

The intersection of machine learning (ML) and product management is a rapidly evolving field. As businesses realize the power of AI-powered solutions across industries like advertising, finance, and healthcare, product managers who understand ML are becoming indispensable. How can product managers build successful careers in this exciting space, grab the unique opportunities ML provides for product innovation, and overcome the challenges of responsibly developing and deploying AI products?

I spoke to Gleb Sinev, an ML product manager with experience at companies like Surfingbird, Handl, and Mantika.ai - a startup that is developing software for the early detection of lung cancer through computer vision technology - who offered valuable insights into the ML product manager's journey. Here's what we can learn from his extensive career.

Adaptability, Diversity, and Understanding

The global ML market was valued at $19.20 billion in 2022 and is expected to grow from $26.03 billion in 2023 to $225.91 billion by 2030. That’s a phenomenal rate of growth and offers an incredible opportunity, but it also shows that ML touches a significant number of business sectors and potential solutions. Sinev suggests that new ML product managers shouldn’t box themselves in by becoming specialists in only one of them until they’ve found their passion.

“Technologies and markets evolve rapidly,” Sinev said. “It is critical to stay updated on ML and associated advancements, such as computer vision, to remain competitive. Because ML can address challenges across industries, from ad personalization to lung cancer detection, product managers should be open to exploring a wide range of use cases.”

Once an ML product manager has been involved in various applications, a passion for one or two of them will arise. Sinev’s advice is never to stop learning about the foundational aspects of ML, but once you have a clear interest in one sector, go as deep as you can to advance your career and come out as a leader in that.“

A foundational understanding of ML is vital at all times, but good product managers benefit even more when they deepen their expertise in specific domains like healthcare or finance once they understand this is the area they love the most,” Sinev said. “This also helps us keep the industry safe. As stewards of powerful technology, ML product managers must ensure data privacy, algorithmic fairness, and transparency, and a deep understanding of the requirements and regulations in one sector is easier to manage than staying broad.”

Harnessing ML to Disrupt Industries

ML's impact on product innovation is undeniable. It has changed how we understand massivamounts of information and allowed us to serve personalized products, services, music, movies, and messaging to everyone with a smartphone or a laptop.

Sinev embarked on his professional journey as a product manager at Surfingbird, an AdTech company housing two distinct platforms: the consumer-centric Surfingbird, which provided personalized "read-more" content, and Relap.io, a B2B native advertising platform.

Later, Sinev joined the startup Dbrain, now rebranded as Handl, as a product and project manager focusing on ML and Computer Vision. Handl later graduated from Y Combinator W20.

Their primary product, DOCR, a SaaS solution for financial institutions, simplifies document information extraction and conversion into editable formats. Under Sinev's leadership, DOCR saw significant success, continually improving its capabilities, particularly in text recognition.

Sinev also spearheaded the development of a data annotation platform, streamlining AI training data labeling across various modalities. He played a key role in product development, team management, and investor and client relations.

“Our platform has been applied in diverse sectors, from food delivery quality control with Dodo Pizza to cattle health monitoring in agriculture, and has been embraced by industry leaders like Nestle, AMG, and Ligolab,” says Sinev.

Importantly, AI excels at analyzing vast medical datasets, from CT scans to genomic data. Tools like Mantika.ai are pioneering ML's role as a vital tool for doctors, potentially saving lives.

As an illustration, algorithms can predict the onset of sepsis up to two days in advance. A significant tool in this field is Google Cloud's "Target and Lead Identification Suite," which was introduced last year. This tool assists companies in forecasting and comprehending protein structures, which is essential for drug design and personalized medicine.

Medical imaging analysis, such as CT scans, MRI, and X-rays, has seen significant advancements thanks to ML and AI. These technologies have demonstrated exceptional proficiency in image interpretation, reaching a level comparable to that of human experts. Notably, ML algorithms have achieved an impressive 87.0% sensitivity and 92.5% specificity.

Sinev’s experience as a lead product manager at Mantika.ai centered around creating a machine learning-based healthcare product from scratch.

“Our team devised a product that could identify potential early-stage lung cancer by flagging suspicious areas in CT scans,” Sinev said. “The product's successful deployment in several clinics reinforced Sinev’s belief in the potential of machine learning technologies for early cancer detection.”

Sinev emphasizes the precision of machine learning in detecting early diseases like cancer, which enhances decision-making in clinical settings. This, in turn, has a significant impact on patient care and treatment planning.

Overcoming Obstacles to ML Adoption

Despite its potential, ML in product development faces challenges.

Trust is paramount, especially in healthcare. Patients and other end-users are understandably cautious about AI's role in this field. Extensive testing, transparency, and human oversight are non-negotiable. But trust is crucial for products in any industry as consumers are more savvy than ever, especially after revelations from whistleblowers who explained how our social media data is being used to manipulate us. This is why ML product managers must stay up-to-date on country-wide, state-wide, and local regulations and laws.

However, those regulations could be holding the industry back. “Outdated regulations sometimes hinder innovation,” Sinev said. “Product managers must work closely with stakeholders and policymakers to develop ethical and practical frameworks for AI deployment and ML applications.”

ML's reliance on data can amplify existing biases, leading to discriminatory outcomes. A robust bias mitigation strategy is crucial. A recent study discovered that an AI tool, trained on medical images such as X-rays and CT scans, had unexpectedly acquired the ability to identify patients' self-reported race. This capability emerged despite the tool being trained solely to aid clinicians in diagnosing patient images.

“The potential misuse of this technology, which can identify patients' race even when their physician cannot, raises concerns about the possibility of future abuse or inadvertently directing subpar care toward communities of color without detection or intervention, so we all have to play a part in ensuring lack of bias,” Sinev said.

The Future of ML Product Managers

ML's role in product development will only expand. Product managers should not only understand ML's technical underpinnings but also drive these considerations.

“While powerful, ML shouldn't be treated as a cure-all,” Sinev said. “Products that integrate ML with the nuanced understanding of human users have the strongest chance of success.”

Product managers should also, according to Sinev, be able to create systems that explain themselves to help eradicate errors, bias, and other problems.

“Increasingly, stakeholders will demand more clarity on how ML systems reach decisions, and not just run with the decisions they make,” Sinev said. “Product managers must collaborate with developers on explainability techniques.”

From social impact to unintended consequences, ML product managers have a broad responsibility. A strong ethical focus is not just good practice, but crucial for long-term success.

According to Sinev, product managers who embrace ML can build exceptional products that solve critical problems innovatively. By remaining adaptable, prioritizing ethics, and fostering collaboration, these professionals will shape the future of countless industries.