OpenVINO is an exceptional open-source toolkit designed to optimize and deploy deep learning models, revolutionizing the performance of vision, audio, and language models across popular frameworks like TensorFlow, PyTorch, and more.
Through iterative testing and collaboration across stakeholder teams, the usability and usefulness of the tool was rescued from abandonment to one of the best examples of SW development at Intel.
Challenge: Enhancing Developer Experience and Integration
In this ambitious project, I took on the role of UX Architect. I also owned the research strategy & design, facing the challenge of improving the developer experience of OpenVINO from discovery to installation and use.
Throughout the course of a year, we conducted eight comprehensive developer experience studies to evaluate the tool, identify pain points, and align it with developer expectations. The research informed the design direction.
As the UX lead, I coordinated with a Technical SME and a UX writer to improve documentation, error avoidance and clarifications. The journey had its share of obstacles, as we encountered initial resistance from stakeholders and engineers due to a lack of UX awareness and limited experience with the process. Additionally, integrating UX into the agile development method posed an extra challenge.
One of our key focus points was understanding the developer journey, which has informed many of our key research deliverables.
Advocating for UX, UI Design and facilitating Collaboration
To overcome these hurdles, a significant portion of my time was dedicated to evangelizing the importance of UX and educating stakeholders and engineering teams on how to effectively collaborate with a UX team.
In scrum meetings, I advocated for UX-related items to be addressed in each sprint, ensuring that user-centric design principles were at the forefront. Working alongside an engineer/developer and an information architect, I led the research and design strategy, orchestrating a harmonious synergy among the team.
One of the artifacts that was useful was the release cadence framework.
Methods: A Holistic Approach to User-Centered Design
We implemented a comprehensive range of methods to gain deep insights into the user experience. Our efforts included conducting UX training sessions for all teams involved, such as stakeholders, product and project managers, content writers, developers, engineers, and marketing and web developers.
Leveraging both internal expertise and managing external vendors, we executed a variety of research techniques, including individual and group user interviews, expert evaluations, journey mapping, moderated and unmoderated usability studies, and triangulation with embedded dynamically adjusted telemetric data.
This holistic approach enabled us to augment qualitative analysis and make well-informed recommendations. We augmented our original understanding of the Developer Experience Journey Map to include more touch points and deeper understanding of how different personas were interacting with the products.
I focused on training cross-functional teams, including stakeholders, product and project managers, developers, and salespeople, on the end-to-end UX process. This comprehensive training covered defining opportunities, creating personas, concept development, market hypothesis testing, and early pilot research.
My guidance resulted in a deep understanding of integrating UX into agile development, including UX architecture, information architecture, content design, and documentation. Our emphasis was on iterative UX assessments, product readiness testing, analysis, and optimization, as well as post-launch evaluations.
By the end of the project, teams were equipped with the skills to deliver user-centric experiences, enhance product quality, and drive business success.
Moderated Qualitative Testing: Gaining Deeper Insights and Evolving Test Approaches
To delve deeper into the user experience of OpenVINO, we conducted multiple moderated qualitative testing sessions each month. As the project progressed, we transitioned to using a fully functional version of the software, enabling us to analyze not only test images and videos but also encourage participants to utilize their own targets from the web. This approach added a layer of authenticity and real-world relevance to our testing process.
Furthermore, to ensure a comprehensive understanding of the user experience, we expanded our testing efforts to include sessions that encompassed Mac, Windows and Linux operating systems. Additionally, we incorporated acceleration hardware into later sessions, providing a more holistic view of the performance and usability of OpenVINO across different platforms and hardware configurations. This broader scope of testing enabled us to address specific challenges and optimize the product for a diverse user base.
Through moderated qualitative testing, we were able to gather valuable feedback, identify pain points, and make informed recommendations for improvement. By continuously adapting our test approaches and embracing the evolving landscape of user needs, we ensured that OpenVINO's user experience was refined and optimized for various operating systems and hardware configurations.
Impact: From Abandonment to Success
Through iterative improvements, OpenVINO witnessed a transformation. Over the course of a year, the tool's satisfaction score increased from 7% to 91%. The number of users surged exponentially, growing from hundreds to tens of thousands, largely driven by positive word-of-mouth and online acclaim. Our success extended beyond product improvement, as we seamlessly integrated the DX (Developer Experience) process and best practices into product development. This accomplishment positioned our team as an example to other software development teams within Intel, showcasing how UX services could be effectively incorporated into their own projects.
By placing the user at the center of our development process and passionately advocating for UX, we not only transformed OpenVINO into a successful product but also paved the way for a more user-centric approach to software development at Intel.