Right from generative AI models to hybrid cloud deployments, HPCL is marrying innovation with pragmatism to tackle modern-day digital challenges of data security, scalability, and cost optimization.
Jayant Gupta, Executive Director of Information Systems at HPCL, has an exclusive conversation- talking about the company’s strategies for leveraging generative AI, managing hybrid cloud environments, and ensuring seamless IT operations.
He reveals how AI is transforming processes in finance, recruitment, and customer interaction while underlining how HPCL has been proactive in security, preventing vendor lock-in, and providing its employees with up-skilling to gain an upper edge in the digital race.
How is HPCL leveraging generative AI, and how do you address the associated security challenges?
What we’ve done with Gen AI is that, at this point in time, we are using it more as models that are enterprise-based and on-prem. We use a couple of models like LLaMA and others to handle the generative part, like creating summaries or building intelligence from there. We’ve been able to apply this across multiple use cases in various streams.
• Finance: Generative AI is used to summarize audit reports, identify commonalities, and streamline processes, enabling auditors to detect recurring issues more efficiently.
• Interactive Bots: Generative AI enhances chatbot capabilities, improving user interactions.
• Recruitment: AI extracts data from scanned documents and forms, automating information entry and ensuring data consistency.
So far, using these on-prem solutions has worked well for us. Since we run everything within our environment, no data—including metadata—is pushed externally. However, we are grappling with challenges around certain LLMs that are cloud-based, such as Microsoft’s CoPilot. While piloting these solutions, we’ve noticed that metadata is sent out, even though the rest of the data remains internal. How safe this is remains an open question for us.
Hindustan Petroleum Corporation Limited (HPCL) has adopted a hybrid cloud model, using on-premises solutions for sensitive data and a public cloud for specialized APIs or functionalities.
Why did HPCL choose a hybrid cloud model, and how do you decide workloads for on-premise versus public cloud?
The decision to adopt a hybrid model is driven by cost considerations and specific functionality needs. With a Tier-4 green enterprise data centre, we prioritize using in-house resources but leverage the public cloud for specialized APIs or functionalities unavailable internally.
This setup allows us to use internal resources extensively, and we only look outside when necessary. Our solutions span across private, public, and hybrid cloud models, depending on the requirements.
When deciding on these solutions, we consider two key factors:
1. Cost: We prioritize more cost-effective solutions.
2. Functionality: If a specific functionality, solution, or API is available on the cloud and is more effective, we adopt it.
With a Tier-4 green enterprise data centre, we prioritize using in-house resources but leverage the public cloud for specialized APIs or functionalities unavailable internally.
In practice, workloads purely consumed and produced by the public are typically best suited for the public cloud. For workloads generated internally but accessed externally, we move them to the public cloud for better access. It ultimately depends on the type of solution required. However, for public-facing solutions, the public cloud is often the preferred choice.
What challenges did you face while managing hybrid cloud environments?
When data is moved to the public cloud, it’s primarily used for access, so it’s not a significant challenge. However, the real challenge arises in terms of response times between internal and external use. Public cloud responses are generally fast when accessed from the public side, but when coming from the enterprise side, they tend to be sluggish. There are noticeable variations in response times between these two environments.
Working with vendors also presents challenges. You’ll always encounter issues with them, but the real limitation lies in the depth of the solutions they provide. Vendors often lack experts who fully understand the specific systems and environments you are using. As a result, it takes time for them to grasp your unique environment and needs.
How does HPCL address vendor lock-in and ensure cost efficiency in cloud solutions?
Vendor lock-in is an inevitable challenge for certain critical solutions like Microsoft Office and SAP. To mitigate this, we adopt a mixed technology stack, leveraging open-source products wherever feasible to reduce costs and maintain flexibility. This strategy minimizes dependencies while optimizing expenses.
How does HPCL ensure data security and compliance across its infrastructure?
We have implemented a Zero Trust Architecture to enhance data security across platforms. Additionally, our CISO team actively monitors compliance requirements, ensuring all systems adhere to stringent security protocols.
How do you improve the resilience and flexibility of HPCL’s IT infrastructure?
We focus on redundant architecture to minimize downtime. Both of our data centres are configured in High Availability (HA) mode, ensuring uninterrupted service. In the event of one data centre being unavailable, the other takes over seamlessly.
What advancements in hybrid cloud technology would most benefit HPCL?
For our enterprise, we’ve observed that there aren’t many solutions available right now that span seamlessly across to the cloud. Our data growth is relatively stable and linear, not exponential, making our data and compute requirements predictable. As such, we don’t anticipate sudden spikes that would necessitate the use of the cloud.
However, given the procurement cycle involved, we do need cloud resources available during periods when we’re unable to deploy our own infrastructure. Currently, not all solutions offer this flexibility. Ideally, we would prefer a solution that allows our on-prem infrastructure to seamlessly extend to the cloud.
inter
How does HPCL stay ahead of advancements in AI and emerging technologies?
We emphasize upskilling and collaboration. Our team participates in technical and strategic courses through partnerships with leading institutes like IIMs and IITs. Additionally, we adopt an “early adopter” approach—piloting emerging technologies, evaluating their feasibility, and swiftly discarding those that do not align with our goals.
Do you have metrics to evaluate the success of new technology implementations?
Yes, we have established clear success criteria for technology adoption. Solutions go through a rigorous evaluation process, and only those that meet our benchmarks are deployed.
This structured approach ensures we derive maximum value from technology investments while staying agile in a rapidly evolving landscape.
Jayant Gupta
Executive Director of Information Systems, Hindustan Petroleum Corporation Limited
aanchalg@cybermedia.co.in
