[Sources] - Gurugram DSS screenshot: local reviewed technical deck asset, 2026-07-28. - AIReSQ, IITGN and FloodReSQ logos: local project assets.

Open with urban inundation as a coupled infrastructure problem. The point is not that every city floods identically, but that dense interdependence turns local water accumulation into network-wide disruption. [Sources] - AP, “Record rains bring Mumbai to a standstill, in photos,” 20 August 2025. https://apnews.com/article/6b1198e8cce8350de7b9875001dab676 - Reuters/WaterAid coverage of flood-drought whiplash across major cities, 12 March 2025. https://www.investing.com/news/world-news/cities-face-whiplash-of-floods-droughts-as-temperatures-rise-study-warns-3922645

Use the sequence to show recurring disruption across different urban forms. These are article summaries, not claims of equal causation or severity. [Sources] - Hindustan Times, 3 September 2025. https://www.hindustantimes.com/cities/gurugram-news/recovering-from-deluge-a-calm-in-gurugram-day-after-chaos-101756838929013.html - AP, 20 August 2025. https://apnews.com/article/6b1198e8cce8350de7b9875001dab676 - Times of India, 7 August 2026. https://timesofindia.indiatimes.com/city/gurgaon/gurgaon-rain-chaos-offices-shift-to-work-from-home-as-flooded-roads-cripple-traffic/articleshow/133021781.cms - AP, 15 July 2025. https://apnews.com/article/09b0e611a1dead720168b3acccc4aeb1

Veracity here means the reliability, provenance and time-validity of heterogeneous evidence. Avoid presenting it as a slogan; explain the operational consequence of stale or mismatched inputs. [Sources] - Gurugram reviewed citywide model and operational event packages, local project records. - UN-Habitat, World Cities Report 2024, infrastructure-disruption discussion. https://unhabitat.org/wcr/

This is a statement about integration and operating tempo, not a dismissal of forecasting, hydraulic consulting or municipal field systems. The enterprise builds on all of them. [Sources] - Synthesis of the Gurugram and Vijayawada deployment workflows and municipal information requirements.

This ERL study linked extreme precipitation, floods, landslides, debris flow and road-network functionality. It established the intellectual move from isolated hazard layers to concurrent physical processes and infrastructure consequences. [Sources] - Dave, Subramanian and Bhatia (2021), Environmental Research Letters 16, 104050. https://doi.org/10.1088/1748-9326/ac2d67

[Sources] - FloodAstra engineering architecture and research lineage, local technical documentation. - Dave, Subramanian and Bhatia (2021), Environmental Research Letters 16, 104050. https://doi.org/10.1088/1748-9326/ac2d67 - Nature Cities (2025), “Unequal urban flood protection can magnify social inequality.” https://www.nature.com/articles/s44284-025-00299-7

[Sources] - Nature Cities (2025), “Unequal urban flood protection can magnify social inequality.” https://www.nature.com/articles/s44284-025-00299-7

The paper reports reduced aggregate losses alongside increased inequality metrics and earlier downstream flooding in some locations. Explain the exact metrics verbally and avoid implying that every intervention redistributes risk. [Sources] - Nature Cities (2025). https://www.nature.com/articles/s44284-025-00299-7 - Indian Express coverage. https://indianexpress.com/article/cities/ahmedabad/study-partial-flood-defences-surat-shifted-risk-vulnerable-communities/

[Sources] - The Better India, “This IIT Team Built an AI System That Can Predict Floods Before the First Drop Falls,” 4 August 2026. https://thebetterindia.com/innovation/iit-gandhinagar-arc-centre-ai-urban-flooding-udit-bhatia-indian-cities-monsoon-12227116 - ARC Centre photographs from the institutional ARC Centre science-to-decision deck, 2026. - IIT Gandhinagar ARC Centre launch. https://news.iitgn.ac.in/iit-gandhinagar-launches-ai-resilience-and-command-centre-for-data-driven-climate-risk-management/ - IIT Gandhinagar, “Beyond the Paper: How AIResQ ClimSols is Protecting Urban India from Flooding.” https://news.iitgn.ac.in/beyond-the-paper-how-airesq-climsols-is-protecting-urban-india-from-flooding/

[Sources] - Rain2Resilience end-to-end intelligence pipeline, local AIReSQ/ARC programme material.

The cricket ground makes the decision problem tangible: rainfall is an input, but recovery time is the operational output. [Sources] - IIT Gandhinagar CampusAstra cricket-ground time-to-play animation, local project output. - IIT Gandhinagar CampusAstra campus-scale maximum-depth map over orthomosaic, local project output.

Use these figures only with the reviewed event definition from the commissioner deck. They frame the storage and conveyance problem; they are not a universal capacity statement. [Sources] - MCG Rain to Resilience commissioner deck, slide 1, local file.

[Sources] - Vijayawada and Gurugram local model-input products.

FloodTwin is the spatial decision interface. The reference product view combines a time-controlled flood surface with critical assets and road accessibility; quantitative claims remain tied to reviewed FloodAstra products. [Sources] - FloodTwin product interface, local AIReSQ programme material. - MCG Rain to Resilience commissioner deck and reviewed Gurugram products.

Do not imply that learned components replace conservation. The novelty is the domain-governed integration and accelerated intervention search. [Sources] - FloodAstra engineering architecture, local source and solver documentation.

This category comparison is conceptual. The internal backup slide names adjacent products and should not be circulated without a fresh feature review. [Sources] - Public product positioning and local FloodAstra capabilities, reviewed August 2026.

Use only the reconciled 3:56 benchmark in the main talk. A separate 124-second engineering run used a different workload and is documented in backup. [Sources] - Gurugram commissioner deck and local benchmark records.

[Sources] - Gurugram terrain audit, solver mass-balance diagnostics and field-validation products, local project outputs.

[Sources] - MCG Rain to Resilience commissioner deck, local file. - Hindustan Times, Gurugram pilot announcement. https://www.hindustantimes.com/cities/gurugram-news/gurugram-to-pilot-flood-management-system-with-iit-gandhinagar-101769797097055.html

This is a reviewed 2025 event product from the V2 evidence library. Keep the event label visible; it is not a live nowcast. [Sources] - Gurugram final-review event animation: July 2025, 133 mm / 12 h, local product.

The 22 August example is deliberately a low-rainfall operational day. It shows that the system suppresses a citywide warning while retaining targeted field checks where local terrain can hold water. [Sources] - FloodAstra operational package 20260822T013505Z_tomorrow_io_gurugram_20260822_0730, QA-passed. - Tomorrow.io Gurugram forecast issued 22 August 2026 at 07:30 IST.

The 96.5% figure must always be presented with the exact matching definition from the commissioner deck. The sensor count refers to the reported 40-sensor programme, not necessarily 40 live feeds at one instant. [Sources] - MCG commissioner deck, slide 5. - Indian Express sensor coverage. https://indianexpress.com/article/cities/delhi/gurgaon-gurugram-ai-sewer-sensors-monsoon-waterlogging-prevention-10726957/ - Hindustan Times sensor coverage. https://www.hindustantimes.com/cities/gurugram-news/mcg-to-install-40-sensors-across-gurugram-to-strengthen-flood-prediction-101780108928844-amp.html

[Sources] - MCG Rain to Resilience commissioner deck, slide 4. - Gurugram ward-report products, local reviewed outputs.

This is the actual first page of the Ward 21 report generated from the selected 22 August 2026 nowcast package. It includes event lineage, ward assignment, model window, localities, action and field-use status. [Sources] - MCG Ward 21 Flood Hotspot Report, selected Flood Watch run 20260822T013505Z_tomorrow_io_gurugram_20260822_0730.

The playable recording is captured from the actual local Three.js viewer used in the commissioner material. It changes season and layer mode while rotating the scene; it is not a simulated decorative animation. [Sources] - MCG commissioner deck, slide 3. - Local groundwater-context 3D viewer and NAQUIM screening products.

The animation is recorded from the operational RechargeTwin using the reviewed July 2025 rainfall, 05 June 2026 municipal assets and the same coupled 2D-surface/1D-drain physics in every configuration. The preferred passing case uses two registered MCG parks, four 50 HP pumps in total, 14,000 m3 finite detention and 12 provisional screened wells. Pumping controls street recession; detention protects the event peak; the 96 L/s recharge screen restores storage over the following day or two. The park well fields are represented as independent investigation zones because lateral aquifer continuity has not been established. [Sources] - FloodAstra D Block two-park optimization summary, local reviewed output, 25 August 2026. - MCG registered park polygons H.No-1977 and H.No-1877; 05 June 2026 asset package. - RechargeTwin operational viewer, local V3 Groundwater workspace.

Do not confuse the Top 10 ranked interventions with the full hotspot population. Scenario Lab begins with the full reviewed set, then ranks candidates. [Sources] - Gurugram Scenario Lab V1/V2 reviewed interface and hotspot catalogue, local outputs.

The playable sequence moves through the actual reviewed interfaces and evidence: full hotspot selection, intervention design, paired GPU comparison and recovery evidence. Defaults are derived from mapped assets and hydraulic context, remain editable, and never enter a run when blocked. [Sources] - Gurugram Scenario Lab reviewed interface and engineering contracts, local outputs. - MCG commissioner deck, slide 2, and Rajendra Ward 34 paired-run products.

Rajendra Ward 34 is a reviewed comparison example. Explain that improvement is multidimensional: reduced surface flooding can coexist with increased return at manholes, requiring a guardrail or redesign. [Sources] - Rajendra Ward 34 completed comparison product, local reviewed output.

[Sources] - Gurugram terrain, topology, intervention and recession audit products, local files.

Keep claims limited to demonstrated network-disruption and rerouting products. Do not imply a live logistics deployment unless verified. [Sources] - MobiReSQ local network and route evidence products.

[Sources] - Vijayawada RTGS Rain2Resilience brief, slide 3, local file.

Do not compare the 73.4-second Vijayawada workload directly with the Gurugram 3:56 benchmark without explaining domain, horizon, mesh and coupling differences. [Sources] - Vijayawada RTGS Rain2Resilience brief, slide 7, local file.

Gurugram and Vijayawada are the primary demonstrated cases in this talk. Ahmedabad and Gandhinagar are shown as scale-out pathways; verify their current contract and deployment status before external delivery. [Sources] - Local Rain2Resilience programme materials and deployment records.

[Sources] - The Better India, “This IIT Team Built an AI System That Can Predict Floods Before the First Drop Falls,” 4 August 2026. https://thebetterindia.com/innovation/iit-gandhinagar-arc-centre-ai-urban-flooding-udit-bhatia-indian-cities-monsoon-12227116 - IIT Gandhinagar AIReSQ feature. https://news.iitgn.ac.in/beyond-the-paper-how-airesq-climsols-is-protecting-urban-india-from-flooding/ - Northeastern Global News, “How AI Flood Prediction is Keeping Highways in India Open.” https://news.northeastern.edu/2026/04/08/ai-flood-prediction-india-monsoons/

Dubai is a proposed application, not a completed deployment. Real-estate, supply-chain and industrial use cases should be described as product extensions unless supported by contracts or validation. [Sources] - Local AIReSQ application visuals and Dubai concept materials.

[Sources] - Synthesis of the Rain2Resilience architecture demonstrated throughout this deck.