Photo by Chris Chow on Unsplash
In our ongoing interview series, Intel VP and GM for Open Ecosystem Arun Gupta hosts live conversations with changemakers in the open source community. Follow him on LinkedIn to tune into the next conversation.
Gupta recently talked with Rafee Tarafdar, CTO of Infosys*, a global leader in next-generation consulting and digital services serving clients in more than 55 countries. Tarafdar has been at Infosys for almost 19 years. He focuses on driving tech strategy, leading clients through transformation journeys, and building tech talent density within the company. Tarafdar also drives Infosys’s transformation from a digital- and cloud-first company to an AI-first company.
The two discussed how Infosys leverages open source technology in large programs, how it decides to invest in emerging tech, and what strategies are in place to help Infosys contribute to open source communities.
This conversation has been edited and condensed for brevity and clarity. Watch the full interview on YouTube*.
Blending commercial and open source tools
Arun Gupta: How does Infosys balance the use of open source software with your own proprietary technologies?
Rafee Tarafdar: As an open source–first company, we always look to use open source technologies where they fit, from software development kits (SDKs) to databases to streaming software. However, we find that even when open source tooling is available, many times integrating them can be challenging—or we may need to drive orchestration across the entire life cycle and the tools aren’t able to fully integrate. In these cases, we may need to make additional investments to bring the orchestration across the different tools. Our goal is always to innovate on top of open source rather than try to replicate what an open source solution already does. We also look to use third-party technology from our partners. Being open source doesn’t mean we won’t use commercial software or SaaS and PaaS software. We choose the tool best suited for our needs and our clients’ needs.
Arun Gupta: When there are issues with open source projects, you typically have to submit a request and wait for the community to address it. How do you balance open source challenges while meeting client timelines?
Rafee Tarafdar: We vet and evaluate open source tools before adopting them. We start by looking at security-related issues and the licensing model and how permissive it is. We evaluate the strength of the community, including how much talent or expertise is backing the project, which plays an important role in innovation. We also evaluate the frequency of commits to help us understand how up to date the project is. We then run the software through a set of checks to ensure it meets the criteria and guidelines we’ve set. This analysis helps us identify risks.
It’s also important to understand what type of open source tool you’re working with. There are two types. Frameworks or SDKs that are used during development, such as frameworks for Angular* or Java* spring boot, often don’t require a high level of support from a production environment standpoint, whereas other tools do, including databases, caches, messaging platforms, and orchestration engines. For these, we look for vendors that provide a premium level of resiliency, scale, and 24/7 support.
When working with proven open source tools, we don’t often have trouble meeting the speed requirements of our clients. For new open source solutions with limited community or vendor support, Infosys works with key stakeholders to ensure we understand the risks and how to manage them. The good news is, with the number of cloud service providers (CSPs) today, there are many managed open source software options available. For instance, when you need speed, choosing a Kubernetes*-managed database means you’ll have support from someone that can take care of some issues in the cloud—and that Infosys can stay true from an open source and neutrality standpoint.
Determining where and when to invest
Arun Gupta: What strategies has Infosys adopted to navigate the evolving landscape of open source technology and to drive sustained innovation?
Rafee Tarafdar: Infosys has specialist communities that focus on 11 key technology areas, including API microservices, AI, and cloud. The communities track developments in these spaces, such as patterns, best practices, and new open source technologies. This information comes from the open source community, the partner community, the startup community, the research community, and our clients. We plot all of these developments along three horizons: horizon one, areas that lack tech investments by clients; horizon two, areas that are currently receiving tech investments; and horizon three, areas that are likely to receive tech investments in the next six to 18 months. If horizon two looks to dominate demand for the foreseeable future, we know to invest in key tooling, partners, vendors, and skills in this area. We evaluate the maturity and stability of technologies in horizon three to help us determine when to invest. Sometimes we’ll invest in technologies that aren’t mature yet because we believe in the potential, like Kafka*. So everything is driven by our TechCompass* and our three horizons.
Arun Gupta: You have a full machine to make this system work, and without it, you wouldn’t be able to keep up with the rapidly evolving landscape.
Rafee Tarafdar: We’re finding that in the AI space, for example, there’s a new open source large language model (LLM) or framework every other week. It’s like a treadmill, and you’re always trying to keep up. The only way we can track everything is through a distributed, community-based model. It’s not possible with a centralized model.
Running open source at the largest scale: “100 million people pay taxes on these platforms”
Arun Gupta: Can you share examples of when Infosys implemented open source stacks for large programs?
Rafee Tarafdar: Infosys runs some of the largest tax platforms for the Indian government. They feature some propriety software, but they largely run on open source components. These platforms support more than 100 million people. In any taxation system, the largest volume of payments comes in on the last day, creating an event similar to Black Friday. We’ve been able to use open source platforms to handle that kind of scale and throughput. We also use open source software in many of the transformation programs that we run for global clients that span across many countries. In fact, many kinds of codebases today run largely on open source, including containers, databases, web servers, messaging, data engineering, and AI. It’s fair to say open source is embedded into everything we do in transformation programs.
Strategically contributing to the open source community
Arun Gupta: In addition to the open source strategy we’ve been talking about, Infosys has also set up an open source program office (OSPO). What’s the purpose of OSPO, and how does it help you promote an open source culture at Infosys?
Rafee Tarafdar: We’ve been a strong consumer of open source, but I don’t think we’ve done a good job of contributing to the community. We established an OSPO to help us become a contributor in a more intentional and strategic way. The first goal of OSPO is to continue to evangelize open source consumption and help build the ecosystem. Second, we wanted to create the playbooks and processes that will help us be an active contributor to the open source community. The third goal is to guide the way we participate in leading open source forums, which will help us learn about innovations at a much closer level. That’s why we’ve partnered with the Cloud Native Computing Foundation (CNCF)* and Linux Foundation Networking (LFN)*. Part of this goal is to take the enterprise innovation from our customers back to the open source community. The OSPO team helps us track progress on these goals. It also helps us formalize standards, processes, and policies so we can make them available to the whole organization.
Arun Gupta: Can you give us some examples of how Infosys contributes to the community?
Rafee Tarafdar: First, if we think an internal project or platform will benefit the larger open source community, we share it on GitHub* so others can use it. We did this for our DevSecOps platform, a platform we created that orchestrates a number of different open source rules to build automation across the entire life cycle. Second, volunteers in our tech community are constantly fixing issues in the open source projects Infosys uses. Infosys has 250 contributors who have made about 3,700 contributions across more than 300 repositories. The third level is contributing to documentation, adoption, and community events. Infosys has beautiful campuses in India and around the world, and we’re always looking to bring the community to our campuses to discuss software issues. Fourth, we look to leverage our partnerships to help shape emerging open source projects. Last, we’re slowly starting to get into protocol definition. We’re involved in defining the specs and protocols for population-scale projects, such as the digital identify platform MOSIP*.
Keeping an eye on emerging tech
Arun Gupta: Can you discuss any trends or emerging technologies in the open source space that Infosys finds particularly exciting?
Rafee Tarafdar: Beyond innovation in containers, there’s a lot of interesting innovation happening in the AI ecosystem, such as LLMs and frameworks like LangChain* and ToolLLM*. We also think WebAssembly* has the potential to offer similar value to Java*. There’s also a lot of innovation happening in applications in media, entertainment, and Internet of Things (IoT) devices.
We’re also watching the entire green computing space. Innovations in how we measure, monitor, and track emissions will help ensure we’re using resources responsibly. More importantly, we need to shift the entire engineering life cycle to focus on enabling sustainable upstream choices such as running workloads on green data centers, choosing the right programming languages, creating efficient models, and helping people track sustainability goals at all levels in the hierarchy. We’re also constantly chasing cybersecurity and privacy improvements and the responsible use of AI. Initially, most of the work was happening on proprietary tools, but now the open source community is driving innovation.
Being the voice of the end user
Arun Gupta: Infosys joined CNCF at the top tier, affording you a governing board seat. What benefit do you see by joining the governing board and how do you plan to influence CNCF strategy?
Rafee Tarafdar: A significant portion of our revenue comes from cloud projects, and we’re doing massive transformation programs for many of our clients. At the heart of our cloud work is the container ecosystem. CNCF is the right forum for us to participate in because that’s where most of the innovation is happening. As we work on transformation programs we’re finding a lot of enterprise-specific challenges. We’ve chosen to take these insights back to the foundation and the community, serving as the voice of the customer to make sure their requirements, needs, and concerns are addressed. Doing so will ultimately help drive further adoption of open source. As we work with the foundation board and different subcommittees to prepare for future investments, we can take the foundation’s road map and start evangelizing it with our clients so that they understand what’s coming. We see our role as bringing together the foundation and Global 2000 companies.
Arun Gupta: Infosys joined CNCF as an end user, and I see Infosys acting on behalf of the end user. You mentioned Infosys is moving from cloud- and digital-first strategy to an AI-first strategy. How does that align with joining CNCF?
Rafee Tarafdar: In early 2022 when we were evaluating where to go next, the foundation’s AI models were becoming fairly powerful and many internal pilots showed potential. The work we’re doing with CNCF around containers, DevSecOps, observability, and telemetry is the foundation of our AI-first strategy. Being AI first is not going to offer any value unless we’re also digital first, cloud first, and data first; those are the three foundational blocks that are going to scale our AI products and platforms.
Arun Gupta: You’re taking an evolutionary approach as opposed to a transformational approach. You’re not completely pivoting to a new direction, but building on top of your expertise in cloud, LLMs, and MLOps.
Rafee Tarafdar: I’m a big proponent of evolutionary architecture. Technology will keep changing, and we need to make sure our architecture, practices, and tooling evolves with it so we can build on it. At some point, if something needs to be replaced, we should be able to do it without significantly impacting the rest of the ecosystem.
Follow Arun Gupta on LinkedIn for more live sessions with changemakers in the open source community.
About the presenter
Arun Gupta, VP and GM of the Open Ecosystem, Intel
Arun Gupta is vice president and general manager of Open Ecosystem Initiatives at Intel Corporation. He has been an open source strategist, advocate, and practitioner for nearly two decades. He has taken companies such as Apple*, Amazon*, and Sun Microsystems* through systemic changes to embrace open source principles, contribute, and collaborate effectively.