From chaos to calm: Inside Tirumala's AI command centre

Hyderabad: Two weeks ago, a jeep driver harassed a pilgrim from Uttar Pradesh on the winding climb to Tirumala’s hilltop shrine in Andhra Pradesh. The devotee, shaken but determined, reached the shrine and filed a police complaint with just the last four digits of the vehicle’s number plate.

Amid thousands of cars going to the temple, authorities traced the driver and vehicle in under 45 minutes. Justice was served swiftly and silently, thanks to a new initiative that watches over every corner of the sacred hill.

The Tirumala’s Sri Venkateswara Temple, one of the richest and most revered Hindu shrines in the world, draws 70,000-80,000 pilgrims every day, with numbers often crossing 1 lakh on festivals and special days. Securing a darshan ticket is considered a prized spiritual opportunity. Yet until recently, the experience was often defined by long queues, fatigue, and the ever- present risk of overcrowding.

Now, the average darshan time on a packed day has dropped from 26 hours to about 12-13 hours, and on normal days, devotees can complete darshan in under two hours, with far less jostling and more peace.

This transformation is powered by the newly-opened AI-enabled Integrated Command Control Centre (ICCC) of the Tirumala Tirupati Devasthanams (TTD), the board that manages the temple complex’s affairs.

TTD Additional Executive Officer Ch Venkaiah Chowdhary told DH: “We want to provide a great spiritual experience, with quick darshan and a peaceful glimpse of the Lord without much jostling. This ICCC has come as a boon.”

Housed in a sleek, compact hub built in just 16 days through a pro bono donor-funded initiative by non-resident Indians, the ICCC processes millions of transactions and data interfaces every hour using powerful NVIDIA-powered video-analytics servers.

Over 6,000 cameras — around 500 of them in sensitive zones — blanket the roughly four square kilometres over which Tirumala is spread. They feed live streams into a unified video management system (VMS).

This system handles about 3.6 lakh payloads per minute, processes 51.2 crore events daily, and generates up to 21,900 crore real-time inferences during peak periods, like the recent Vaikuntha Ekadasi, when around 11 lakh pilgrims visited over 11-12 days.

How it works

At the heart of the operation are the Vaikuntham Queue Complexes I and II — multi‑level, semi‑circular structures designed to manage the flow of devotees with military precision.

Queue Complex‑I, the older facility, caters to Arjita Seva and Special Darshan (paid darshan) pilgrims, while Queue Complex‑II handles Sarvadarsanam (free darshan).

Both complexes have over 60 compartments, each holding about 450 pilgrims, arranged in six-seven lines that move through narrow corridors toward the sanctum.

Inside the ICCC, around 30 workstations operate round the clock. Crowd-counting cameras at 69 queue gates and compartment sections ensure that capacities are never exceeded. If a section crosses its threshold, a red alert prompts TTD staff to adjust the flow, pack compartments more efficiently, or hold back new entrants until space frees up.

This approach has apparently helped keep Tirumala stampede‑free even during record‑breaking days when over 14,000 vehicles and 3,500 from the previous day converged on the hilltop recently.

AI cameras and facial‑recognition systems flag suspicious behaviour — such as pickpocketing or loitering — in real time and compare them against a database of known “bad actors” at key entry points like Alipiri and Srivari Mettu.

“All security cameras are now integrated into this centre, enabling real‑time monitoring of pilgrims, missing persons, vehicles, and lost luggage,” said TTD Chief Vigilance and Security Officer KV Murali Krishna.

“We can scale this up to link with criminal records for enhanced security,” the senior IPS officer told DH.

The origins

The idea for the ICCC was born during IT Minister Nara Lokesh’s 2024 visit to Silicon Valley in the US, where he studied smart‑city projects, AI platforms, and digital‑twin technologies.

“We identified real challenges like operations being siloed and pilgrim crowds being massive despite excellent infrastructure, and there emerged a clear need for AI‑driven solutions,” explained Jaya Prasad Vejendla, a Guntur‑born entrepreneur who spent 27 years in Silicon Valley and served as the project’s chief architect.

“With domain expertise from officers like Venkaiah Chowdhary, we built a business‑application framework for temple governance that boosts effectiveness in darshan management, security, laddu distribution, real‑time situational awareness, pilgrim‑behaviour analysis, and spotting pickpockets or missing persons,” Jaya Prasad told DH.

The technology stack combines intelligence platform Kloudspot’s twin digital‑monitoring solution with NVIDIA’s high‑performance computing infrastructure. Predictive analytics forecast Sarvadarsanam wait times, 3D congestion visualisations map crowd density in real time, and unified cyber‑threat monitoring safeguards both pilgrims and temple systems.