The engineer who saw the ceiling at Meta - and decided to rebuild it for everyone else
Somewhere inside Meta's infrastructure org, circa 2019, Abhishek Choudhary was watching something break in slow motion. Not the kind of break that triggers an incident report - but the subtler kind, where the gap between what AI can do and what enterprises can actually run keeps widening. He'd spent six-plus years helping serve video to over a billion people. He understood the machinery. He also understood exactly why no one else could replicate it.
That's the founding thesis of TrueFoundry, and it's less a pivot story than a very specific frustration arriving at the right historical moment. Choudhary, along with his IIT Kharagpur batchmates Nikunj Bajaj and Anuraag Gupta - friends for over fifteen years before they ever sat in a boardroom together - launched TrueFoundry in 2021 to bring Meta-scale AI automation to organizations that don't have Meta's engineering head count. The pitch is almost insultingly simple: your team shouldn't spend months babysitting deployment pipelines when the models are the actual product.
"At TrueFoundry, we're building a future where AI manages AI - removing the bottlenecks of human intervention and unlocking unparalleled speed and scale."
- Abhishek Choudhary, Co-Founder & CTO, TrueFoundryAs CTO, Choudhary drives TrueFoundry's technical architecture - a platform built for the messy reality of enterprise AI deployment. We're talking multi-cloud, multi-framework, Kubernetes-native infrastructure with built-in model governance, cost tracking, observability, and zero data leakage. The kind of system that took Meta years and hundreds of engineers to build, available to an enterprise team in roughly two months.
The performance numbers are worth examining. TrueFoundry's platform handles 350+ requests per second on a single vCPU, delivers roughly 10 millisecond latency under production load, and requires zero performance tuning to get there. These aren't marketing benchmarks - they're a direct consequence of how Choudhary thinks about infrastructure: as a solved problem that organizations shouldn't have to keep re-solving.
From Bell Labs intern to Senior Staff Engineer at Meta - then straight into the deep end
Choudhary arrived at IIT Kharagpur in 2009, part of a Computer Science batch that would, two decades later, collectively raise $21 million. His first taste of serious research infrastructure came as an intern at Bell Laboratories - the institution that gave the world UNIX and C - followed by time at Avaya Labs. By the time he graduated in 2013, the trajectory was set: distributed systems, at scale, for real.
At Meta (then Facebook), he moved fast. Roles in mobile reliability, video performance, and a stint building Wormhole - Facebook's caching infrastructure, the kind of system where every millisecond is somebody's revenue - prepared him for the specific problem TrueFoundry would later solve. When you've optimized video encoding and CDN performance for a billion concurrent users, the idea of an enterprise spending six months just getting a model into production feels like leaving money on the floor.
His trajectory at Meta wasn't slow: from engineer to Senior Staff Engineer in infrastructure, one of the fastest progressions the company's org chart allows. The title matters less than what it represents - a level where you're not just writing systems, you're setting the direction for how systems are built.
"The future belongs to full-stack developers - creators who can build, test, deploy, and scale compound AI applications independently."
- Abhishek Choudhary, announcing TrueFoundry's Series ABefore TrueFoundry, there was EntHire - an AI-based recruitment platform built by the same trio in 2020. It lasted about a year before being acquired by Info Edge, India's largest job portal group. A clean acquisition, a proof of concept as co-founders, and a useful data point: these three could build something real and sell it. EntHire wasn't a failed experiment. It was a warm-up.
What TrueFoundry actually does - and why enterprises keep signing
TrueFoundry sits at the intersection of three problems that compound each other in the enterprise: deployment complexity, governance requirements, and infrastructure fragmentation. Most organizations trying to run production AI have a dozen cloud accounts, a security team with questions, and a data science team wondering why it takes three months to get a model reviewed and released.
TrueFoundry collapses that gap. The platform is Kubernetes-native - which means it runs on your infrastructure, not theirs - with support for multi-cloud, hybrid, and air-gapped deployments. Model governance, role-based access control, audit logs, cost tracking, and OpenTelemetry-based observability are baked in, not bolted on. For regulated industries - healthcare, finance, defence - this distinction matters enormously.
The customer list reflects it: NVIDIA, Siemens Healthineers, ResMed, Automation Anywhere. These are organizations with compliance requirements that make "just use a managed service" a non-starter. TrueFoundry's answer is a self-hosted platform with the automation of a managed service - the infrastructure equivalent of having it both ways, and actually pulling it off.
By February 2025, when Intel Capital led a $19 million Series A - joined by Peak XV (formerly Sequoia India), Eniac Ventures, Jump Capital, and a roster of angels including Gokul Rajaram, Mohit Aron, and Cyan Banister - TrueFoundry had deployed over 1,000 AI/ML clusters and achieved 4x year-over-year customer growth. The "Agent on Autopilot" vision Choudhary outlined in the funding announcement - AI systems that monitor, maintain, and optimize themselves - positions the company for the next wave of enterprise adoption.
The scorecard
Advanced to Senior Staff Engineer at Meta - among the fastest trajectories in the company's infrastructure org
Co-founded EntHire; acquired by Info Edge before TrueFoundry's Series A
TrueFoundry deployed 1,000+ AI/ML clusters across enterprise clients including NVIDIA and Siemens
Raised $21.3M total - seed from Sequoia Surge, Series A led by Intel Capital
Speaker at DataHack Summit 2023 on "LLM Applications and their Cost Dynamics"
Built production infrastructure at Meta serving video to 1B+ users - directly informing TrueFoundry's architecture
The stack he actually works in
Choudhary's technical writing covers AI gateway architecture, LLM load balancing and observability, MCP gateway integration, and enterprise deployment strategies. He's an active contributor to TrueFoundry's blog - an unusual CTO habit that signals something about how seriously he takes the documentation layer of building developer tools.
$21.3M and a very specific angel list
The investors tell the story as much as the numbers. Intel Capital for enterprise infrastructure credibility. Peak XV (Sequoia India) as a continuing believer since the seed. Jump Capital, Eniac Ventures as fresh conviction. And angels who understand what it means to build developer infrastructure at scale: Gokul Rajaram (DoorDash/Google), Mohit Aron (Nutanix/Cohesity), Ankit Sobti (Postman co-founder), Cyan Banister (early Uber/SpaceX), and Lenny Rachitsky, the product-minded newsletter writer who rarely backs infrastructure plays.