Programmatic advertising has long promised efficiency – auctions running in milliseconds, data-driven pricing, and automated demand. But for publishers, the reality has often been far messier. The modern ad stack spans dozens of intermediaries, multiple demand paths, and highly asymmetric access to data. Industry supply-chain transparency studies have repeatedly shown how difficult it is for publishers to see who bought their inventory, at what price, and through which path, even after the auction is over.

Vijay Kumar has spent more than a decade inside that complexity, building technology to give publishers something the market rarely offers: control.

Kumar is the founder of Mile, an ad-tech company designed to solve what he calls the “last-mile problem” in monetization – the point where auction mechanics, pricing strategy, and operational reality decide what revenue actually lands. The company’s name reflects that dual meaning: going the extra mile for customers, and optimizing the last mile where most yield is won or lost.

When programmatic was still opaque

Kumar’s work began in APAC in the early-to-mid 2010s, when programmatic was still gaining traction and many publishers were monetizing through ad-networks running arbitrage. Publishers were generating revenue, but they had little idea who was buying their impressions, how prices were set, or how much margin was being taken by intermediaries.

That lack of visibility was structural. Without transparent auctions or bidder-level data, publishers couldn’t tell whether they were underpricing inventory, losing demand to latency, or being routed through inefficient supply paths.

Kumar was among the earliest in the region to deploy Prebid-style header bidding as a managed service, giving publishers access to open, competitive auctions rather than closed network waterfalls. But he went further: Mile packaged the entire sell-side into a single, transparent operating layer – auction infrastructure, demand access, yield optimization, and white-glove ad-ops, with economics aligned to the publisher.

One large local news media group made Mile its exclusive programmatic partner. Over the next three years, the publisher tripled its overall programmatic revenue by consolidating its stack onto Mile’s transparent auction and optimization platform, replacing fragmented networks with a unified system that delivered higher efficiency, better demand access, and real visibility into how its inventory was being monetized.

That pattern – replacing opacity with infrastructure, would become Mile’s blueprint.

Data became the real bottleneck

By 2018–2019, as Mile entered the U.S., most enterprise publishers already had header bidding and dozens of SSPs. What they lacked was not tools, it was understanding.

When revenue moved, teams often couldn’t answer why. Which partner changed behavior? Was it a pricing issue? A timeout? A demand pullback? Without log-level and cross-dimensional data, even sophisticated publishers were flying blind.

Kumar led Mile’s pivot toward building a real-time monetization intelligence layer that ingested auction-level telemetry across bidders, geographies, formats, timeouts. Instead of waiting for aggregated reports days later, ad-ops teams could see what was happening while it was happening.

The difference this made became clear in a crisis at a large global gaming publisher. The company saw a sudden, alarming drop in fill rate across its sites, even though it hadn’t changed its ad stack. Using Mile’s cross-dimensional real-time analytics, the publisher was able, in under 30 minutes, to isolate the cause: its top SSP, responsible for roughly 60% of revenue in the U.K. and Germany, had stopped receiving demand from DV360 in those geographies.

What would normally take weeks of forensic reporting and guesswork took minutes. The publisher contacted the SSP, demand was restored, and monetization normalized within hours. The problem was never “the site” – it was one broken demand path. Without Mile’s data infrastructure, it would have remained invisible.

When buyers used AI and sellers still guessed

By the early 2020s, DSPs were already using machine learning to optimize bids in real time. Publishers, by contrast, were still setting floors with static rules, blunt thresholds, or SSP-recommended values – tools that couldn’t react to the dynamics of live auctions.

Kumar believed that if buyers were algorithmic, sellers needed to be too.

In 2022, Mile launched one of the first machine-learning-driven dynamic price floor systems designed specifically for the sell-side. Instead of fixed pricing, Mile’s models learned from real-time bid behavior – win rates, bid density, buyer competition, geo, device patterns, and predicted the optimal floor for every slice of inventory.

For one of the largest digital sports destinations in its category in the world, Mile’s AI-powered dynamic floors delivered an 18% RPM lift during peak season, and continued to improve yield as the system learned. Floors became adaptive rather than static, tuned not by spreadsheets, but by live market behavior.

Crucially, Mile’s system was independent. It didn’t resell inventory, didn’t compete for margin, and didn’t optimize for anything except publisher revenue.

From tools to an operating system

By 2025–2026, Mile had evolved into what Kumar describes as an AI-powered Ad Monetization Operating System, a unified control layer that publishers can deploy modularly or end-to-end.

The platform now combines:

• Real-time telemetry and analytics
• Autonomous optimization agents for experimentations
• Custom machine-learning models for pricing and traffic shaping
• Publisher LLM that surface revenue insights across the stack
• Self-serve UI for ad-management

Mile today helps global enterprise publishers and sales houses manage and optimize tens of billions of ad impressions per month and more than $250 million in annual programmatic revenue, across thousands of sites/domains on the web.

A decade inside the sell-side

Kumar’s credibility doesn’t come from chasing trends. It comes from having built through every layer of the sell-side as it evolved – from ad-networks to header bidding, from header bidding to real-time analytics, from analytics to machine-learning-driven markets.

“Mile didn’t start with AI,” Kumar says. “It started with transparency. Then data. Then automation. Each layer only exists because publishers kept running into the same problem, they needed to understand and control their own monetization.”

That philosophy has turned Mile into something rare in ad tech: not just a tool, but infrastructure publishers depend on to run their business.