Cloud computing has become an unseen drain on business budgets. Companies spend millions on third-party cloud providers, often losing a large share of that to idle servers and underused resources. What was meant to bring flexibility has instead created an efficiency gap few can afford to ignore.

Pump tackles that problem by using machine learning, AI, and advanced statistics to find nuanced and hidden savings opportunities across customers' cloud environments. Its platform applies these models to keep workloads and costs in constant balance.

Leading that effort is founding software engineer Shrivant Bhartia, who built the core systems that power Pump's automation and drive its cloud optimization platform at scale.

The Hidden Cost of the Cloud

For many businesses, the cloud's promise of elasticity comes with a hidden tax. As operations scale, engineers prioritise uptime and performance, often over efficiency. A single server upgrade, from a $1/hour to $1.50/ hour instance, seems harmless at first. But across thousands of servers, that extra fifty cents compounds into thousands of dollars every hour and millions every year.

The result is creeping cost bloat—a stagnating drain that often goes unnoticed until finance teams sound alarms.

Recent studies illustrate just how serious the problem is: up to 32% of cloud spending was identified as wasted due to unused or over-provisioned resources, and a Business Wire report finds 78% of companies estimate between 21 and 50% of their cloud expenditure is lost to waste annually.

As Shrivant Bhartia, founding engineer of Pump, explains, "As companies grow, usage increases. And when you're focused on velocity and feature output, you care less about efficiency in code, so you just get more powerful servers to run it faster. As a result, numbers can go from thousands to millions in months. Suddenly, business stakeholders begin panicking, and engineers then have to undertake a huge effort to reduce cloud bills."

How Pump Automates the Economics of Cloud Spending

This is the problem that Pump set out to fix. In less than two years, the company has grown into a large, predominant player in cloud cost optimization, now serving thousands of businesses across AWS, Google Cloud, and Azure.

The core of Pump is built around advanced analytics, distributed systems, and signal processing algorithms. By aggregating and analyzing usage data from accounts across different cloud services, Pump's platform identifies nuanced cost inefficiencies and hidden savings opportunities that would otherwise go undetected. Using machine learning and predictive modeling, it surfaces discounted capacity options and optimizes resource allocation, delivering enterprise-level savings to businesses of every size.

Once integrated, Pump's system operates within the billing layer of a company's cloud account, analyzing its usage and cost data. Using statistical modeling, predictive analytics, and AI, it can forecast consumption patterns, identify waste, and automatically reconfigure workloads or recommend reserved capacity purchases. And because Pump's automation is layered on top of its core analytics engine, each optimization compounds across the platform, with every new data point sharpening the precision of its models for all customers.

For a large enterprise running large cloud instances, Pump operates like a full-time FinOps engineer, continuously recalibrating costs in real-time. For smaller teams, it acts as an invisible analyst, finding savings they would never have the bandwidth to pursue. Either way, the result is the same: smarter infrastructure that aims to reduce costs without affecting performance.

Shrivant Bhartia: The Driving Force Behind Pump's Architecture

Before joining Pump, Shrivant Bhartia had already built a career defined by early-stage grit and technical depth. As the first engineer at Inhabitr, a company with a valuation in the millions, he worked side-by-side with former McKinsey partners, solving operational problems that blended software and business strategy.

At Pump, Bhartia drives the design and refinement of Pump's optimization engine, the main software layer that continuously analyzes cloud usage data to find potential spending inefficiencies and recommend cost-saving actions. His engineering teams develop the orchestration systems that can execute cloud purchasing decisions (like reserved instance commitments or compute reallocation) safely and automatically.

Much of his recent focus has been on improving Pump's prediction layer, where systems forecast resource consumption weeks in advance, enabling the system to optimize spending before it occurs. He's deeply involved in the platform's reliability engineering, ensuring each automation routine is as auditable and reversible as possible so that customers can trust the system to act without any risk of disruption.

Continuing to Evolve the Engineering Behind Pump's Growth

When Bhartia first began building at Pump, his focus was what he calls an "80/20 mindset: just get it out, get feedback, and move." But as the company kept growing, his approach evolved toward ensuring long–term permanence, which meant developing systems designed to withstand load, maintain reliability, and grow without constant rewrites.

That evolution in Bhartia's engineering work mirrors the transformation of Pump itself. With thousands of customers and rising demand, as well as recognition from accelerators like Y Combinator, the company's engineering challenges have only grown more complex. Bhartia plays an important role in the expansion of Pump's technical organization (expected to reach 100+ engineers by year's end) while re-architecting systems to handle greater automation, reliability, and predictive depth.

"You have to balance speed with sustainability," he says, describing the shift from startup urgency to long-term engineering discipline.

Pump's trajectory captures Bhartia's conviction that the best engineering fades into the background of everyday business. With Shrivant Bhartia driving its technical innovation, the company is building technology that aims to function in the background, quietly reshaping how entire companies manage their spending.