Building and scaling high-performance data science teams can be a complex endeavor with various challenges. Altair’s global AI survey reveals that 58% of organisations experience shortage of talent and the time it takes to upskill current employees is the most prevalent problem in their AI adoption strategy. Diversifying skills within teams, aligning them with the company’s culture, and ensuring cultural fit adds complexity. Data access and quality, scalability, interdisciplinary collaboration, and retaining talent all pose hurdles. Resource allocation remains uncertain, given the unclear ROI. Rapidly evolving technology necessitates continuous learning, while ethical and regulatory compliance requires vigilant oversight.
To address these challenges in diverse settings, BFSI experts from large, global, and small banks, new age fintechs, and NBFCs, convened to share their insights and best practices in a panel discussion organised by Altair and ET Edge, offering valuable takeaways for those navigating the data science labyrinth.
Centre of Excellence (COE) as a Catalyst for Efficiency and Innovation
The establishment of a Centre of Excellence (COE) at Wells Fargo has brought about a transformative shift in their data science operations. The COE’s strategic focus on creating tailored solutions for various business teams spanning different domains is having a profound impact by reducing friction and enhancing operational efficiency. Particularly remarkable is the COE’s ability to address up to 60% of the requirements through repeatable solutions, streamlined processes, and resource optimization. Furthermore, the panel members state that this efficiency is reinforced by a commitment to practices such as team rotation, continuous skills upgradation, and a flexible approach to role allocation, all of which contribute to diversifying skills, staying abreast with evolving technologies, and ensuring that expertise takes precedence over rigid templates.
Go Digit emphasises fostering innovation with the same approach. Innovation thrives in organizations that integrate data science teams within business units. Creating mini-Centres of Excellence for specific solutions promotes creativity. Furthermore, ideas should be given precedence over hierarchy. Additionally, automation of repetitive tasks liberates teams to focus on strategic tasks, striking a balance between tactical and strategic work. The panel encourages setting achievable targets, recognizing that the journey cannot be endless.
Key Measures to Achieve High-Performance
The panel discussed about the framework for success:
Start with “Why”: Understand the purpose and goals of your data science journey.
Data Science teams need to be convinced that the work they are doing is going to add to their experience and bring value to their career. For this, the company needs to have the right data science problems to work on, data to solve the problem, subject matter experts to guide the team, an end-to-end integration team to take it to production, and a culture of collaboration not just among the various specialized data science teams but throughout the organization.
• Business Understanding: Treat data science as an engineering problem and develop a deep understanding of the business. Facilitate a collaborative atmosphere not only across the data science teams but also other teams in the organization working on various transformation products/projects as the data science team needs to work closely with other functionaries.
Psychological Safety: Create an environment where team members feel safe to take risks and innovate.
Curiosity: A fundamental requirement for data scientists to keep learning and growing.
Scaling Success
To achieve scalability in the dynamic field of data science, it’s imperative to adopt a holistic approach. Leaders from Jocata Financial Services, Jana Small Finance Bank, Refyne, Fi Money, Vistaar Financials, and Razorpay emphasised this holistic approach for scalability. It begins with the recruitment of individuals who not only possess strong technical skills but also demonstrate a commitment to rapid learning and adaptability, enabling them to stay relevant in an ever-evolving landscape. To build a full-fledged ML competency, a wide range of skilled experts are essential, including data engineers, data scientists, and analysts with specialized skills in computer vision, NLP, structured data models, subject matter experts, annotators, and a marketing/sales team well-versed in product understanding and sales strategies. Notably, the strategies for achieving this include onboarding resources with a few years’ experience and upskilling them through recognized certifications and bringing in interns from colleges with a basic background in machine learning. Additionally, investing in experienced resources for specialized tasks, and appointing a “broker” who acts as a bridge between sectors and facilitates collaboration within the data science team and with other projects requiring data science integration.
Conclusion
Fostering collaboration and success entails understanding each team member’s unique aspirations, embracing simplicity to streamline processes, promoting cross-pollination of ideas for innovation, and encouraging curiosity among business teams. Ultimately, consistent team behaviour, characterised by adherence to established processes and effective communication, is the linchpin for long-term success and scalability in the field of data science.
(This article is authored by Manmeet Kahlon, Director- Data & Analytics, PWC Moderator)
