New Release
Learn More
Your submission has been received!
Thank you for submitting!
Thank you for submitting!
Download your PDF
Oops! Something went wrong while submitting the form.
Table of Contents
Get weekly insights on modern data delivered to your inbox, straight from our hand-picked curations!
the following is a revised edition.
We added a summarised version below for those who prefer the written word, made easy for you to skim and record top insights! 📝
The insights below are summarised versions of Sheila Torrico’s actual dialogue. Feel free to refer to the transcript or play the audio/video to capture the true essence and details of her as-is insights. There’s also a lot more information and hidden bytes of wonder in the interview, listen in for a treat!
Additional note from community moderators: We’re presenting the insights as-is (and summarised) and do not promote any specific tool, platform, or brand. This initiative is to simply share raw experiences & opinions from actual voices in the analytics space to further discussions.
Prefer watching over listening? Watch the Full Episode here ⚡️
Sheila Torrico is a noted advocate of GenAI and Data Products. With expertise in AI/ML initiatives across diverse industries, Sheila excels in leading global teams, fostering cross-functional collaboration, and delivering innovative solutions.
She has been a founding member of the Data Product Leadership Community as well as donned multiple hats as Project Manager, Evangelist, and Consultant at noted companies such as Google, VMware, Kintsugi, Capgemini, and more. Currently, she's serving as the AI and Data Technical Manager at MyLife.
Sheila focuses on bridging tech-business gaps, fostering cross-functional partnerships, and applying my multidisciplinary experience in AI/ML, UX/UI, and Product Management to drive impactful results. We highly appreciate her joining the MD101 initiative and sharing his much-valued insights with us!
We’ve covered a RANGE of topics with Sheila. Dive in! 🤿
In my experience, data has always been central to decision-making, but the real shift came when I took a sabbatical to become a data scientist. I learned that it's not just about structured data but also unstructured data like text, which many overlook. One project that stood out involved using Natural Language Processing (NLP) to cluster customer feedback and categorize it as product-related, safety concerns, or compliments.
I also worked on a product that used speech to screen for depression, which showed the potential for machine learning in healthcare. My role was to ensure the engineers stayed focused and connected their work to tangible business outcomes, ultimately leading to faster product innovation.
When implementing data products, a critical first step is defining what a data product actually is. Many businesses overlook this and jump into the technical side without understanding the end goal. The key is to understand the data sources, work with the business to define success and align on use cases.
It's essential to assess your existing data and infrastructure before diving into machine learning or data science initiatives. A data product manager plays a crucial role in bridging the business and technical teams, ensuring that everyone is aligned and can work together toward successful implementation.
My approach is to be a connector between the business and data teams, ensuring communication flows smoothly. I leverage my experience as both a consultant and data scientist to foresee potential risks and help align business goals with data outputs.
For example, when a business wants to sell more cars, I work closely with teams to translate technical outputs, like accuracy percentages, into actionable business insights. It's all about making sure the right data and outcomes align to meet the business's goals and managing risks, including ensuring the necessary data is available.
Data governance frameworks must be tailored to the organization's specific needs, acknowledging that every company has unique systems and resources. The key is creating the right foundational guardrails, focusing on the people who advocate and enforce the governance. Data governance isn’t just about restricting access; it’s about ensuring the right people can access the data. The success of any governance framework ultimately relies on the human element, with data governance ambassadors playing a pivotal role in its effectiveness.
In smaller settings, like startups, people must be flexible and often handle multiple roles. You'll have data engineers, machine learning engineers, and even product managers contributing to the data stack. As the organization grows, roles become more specialized—data architects design the data infrastructure, while data engineers manage data quality and transformations for the data science teams.
Machine learning engineers take the models from data scientists and deploy them. It’s important not to forget the upstream and downstream stakeholders who produce and consume this data, as they play a vital role in the process.
Understanding core concepts in data is crucial over focusing on specific tools. Master SQL and Python. SQL helps with data manipulation, and Python is essential for data analysis and machine learning. Being well-rounded in these fundamentals will help you communicate across roles and grow in the field.
When exploring tools, consider their UX/UI, functionality, and how they fit your team's needs. Tools like Phoenix allow real-time evaluation of LLMs, while others like weights and biases help track and visualize ML experiments. User-centric design and easy communication of results are key for effective adoption.
The key to overcoming challenges with these tools lies in understanding the goal. Snowflake and DBT have strengths, but developers must assess the tools based on team goals and target outcomes. Success requires a mix of foundational skills, knowledge, and tools.
Growth isn't just about metrics like user adoption; it’s about understanding the business’s true goals. Sometimes, thinking outside the box and redefining metrics like faster product innovation can yield better outcomes.
Key missing pieces include clear data governance standards, stronger data security measures, and more clean or unseen data for models. Data governance and security need more attention to improve the stack's reliability.
The concept of semantic modelling and semantic data products is still evolving. The definitions for these terms are not fully agreed upon within the community. While solutions are being developed, there's a need for a deeper consensus on terms like "data product" and "semantic layer" across organizations. Semantic layers are complex, and there's an ongoing effort to simplify their understanding, as discussed by Brian O'Neill.
Data teams are becoming better at aligning their efforts with business KPIs like ROI and ARR. Many businesses have already implemented data initiatives like AI or self-service analytics and are now refining their ROI metrics. These businesses are comfortable adjusting KPIs, such as model accuracy, to better reflect their goals, which fosters a stronger connection between data teams and business outcomes.
Data catalogs can help reduce the overwhelm of downstream requests, but they are most effective when combined with human expertise. Having knowledgeable people in the loop who understand the data and requirements makes a significant difference in shaping the right models. The combination of data catalogs and human input streamlines processes and ensures quality outputs.
Future-proofing data infrastructure is a challenge, but it can be managed by working closely with stakeholders like the CTO, business leaders, and chief data officers. While new technology offers exciting possibilities, it's essential to evaluate whether it aligns with the existing architecture. Building modular, foundational infrastructure enables flexibility to scale and absorb new tech, but caution is needed to avoid blindly chasing shiny new tools.
Being part of the Data Product Leadership Community (DPLC) has provided valuable exposure to the global evolution of data products. While there’s still much work to do, the community fosters shared language and understanding about data products across regions. The DPLC discussions, especially those led by Brian O'Neill, are crucial for building consensus and progressing the field.
To understand how to use a data product, an analytics engineer must collaborate with the data product manager. This manager will guide them on why the product makes sense, focusing on user needs and feedback. Key conversations should involve the use case and the tech stack, ensuring the engineer aligns with the product manager to effectively integrate the data product.
A good data product is something an engineer transforms into a clean data asset. For a product like a self-service dashboard, the user’s feedback is crucial to ensure it meets expectations. The same applies to more complex ML or AI products, where collaboration with the ML team is essential to refine outputs and ensure the product delivers real value.
The key to evolving a data product is actively seeking user feedback. Whether from external or internal users, feedback from interviews or informal chats helps refine the product. Iteration is crucial, and even a proof of concept can offer valuable insights for improvement, making the product more usable and impactful.
To boost adoption, understanding user pain points is key. Don’t wait for a perfect product; launch a functional proof of concept early. Actively seek user feedback, incorporate it, and improve iteratively. This approach shows users their feedback is valued, making them more likely to engage with and adopt the product.
To stay updated, I rely on the DPLC community and follow industry leaders like Sarah Floris, Chad Sanderson, Joe Reis, Gabriela de Queiroz, and Nick Vidion. They offer hands-on insights, especially in data governance and machine learning. Foundational books in data engineering and machine learning are also essential for staying grounded in the basics.
Maintaining work-life balance involves setting boundaries despite the temptation to take on exciting new projects. It's important to recognize your capacity and prioritize, especially in high-demand roles. While I love working with data and collaborating with teams, I strive to balance my professional passion with personal boundaries.
📝 Note from Editor
The above insights are summarised versions of Sheila Torrico’s actual dialogue. Feel free to refer to the transcript or play the audio/video to capture the true essence and details of her as-is insights. There’s also a lot more information and hidden bytes of wonder in the interview, listen in for a treat!
Thanks for reading Modern Data 101! Subscribe for free to receive new posts and support our work.
Connect with me on LinkedIn 🙌🏻
