Matthew Kirkham
Bountiful
Turn every market in to a shop
VAST DataNVIDIACoreWeave (Weights & Biases)
View project Real-Time Video Agents Hack
Project gallery
Explore the projects and technologies built at the hackathon.
Showing 1–30 of 41 projects
Emily Yuan
Sentiment Engine
A real-time video agent that pairs on-screen game events with players' facial reactions and makes hours of gameplay searchable in plain language.
VAST DataCoreWeave (Weights & Biases)NVIDIA
View project 
jayakumaronline
Vision Zero Agent
Real-time alerts for news desks facing hours of raw footage: NVIDIA Cosmos checks every 5 s segment as it arrives for the event you describe (explosions: 51 of 56 found), and the same agent flags close calls on street and warehouse cameras.
VAST DataSpaceXAI (Cursor)CoreWeave (Weights & Biases)NVIDIA
View project Jeremiah Richard
RehabNinja
AI enabled rehab game
NVIDIASpaceXAI (Cursor)
View project Shreeraj P
StreetAudit
Finds which streets need repainting of road markings
CoreWeave (Weights & Biases)
View project 
katyayanadri02
Scenetrace
SceneTrace is an AI-powered video safety investigation platform that helps cyclists, road safety teams, and urban planners identify potential near-misses, investigate contributing factors, and generate evidence-backed accident prevention recommendations from real-world footage.
VAST DataSpaceXAI (Cursor)NVIDIA
View project 
Jamie Roberts
wahz
wahz is a musical performance and teaching agent.
VAST DataCoreWeave (Weights & Biases)NVIDIA
View project 
Shay Bouchles
AisleGuard
AisleGuard is an autonomous warehouse safety agent that detects worker–forklift proximity risks, searches historical video for similar incidents, and triggers alerts when recurring hazards emerge.
VAST DataSpaceXAI (Cursor)CoreWeave (Weights & Biases)NVIDIA
View project jirubin159
Visual Actions
It's a menu bar app that controls the desktop with gestures, using modes the way Vim does. You raise an open palm to "arm" it, then make a command gesture:
VAST DataCoreWeave (Weights & Biases)NVIDIA
View project Soulemane Sow
Sightline
Self-configuring video agents that understand any feed, decide what matters, and automatically monitor, investigate, and mark important moments.
VAST DataSpaceXAI (Cursor)CoreWeave (Weights & Biases)NVIDIA
View project 
Avinash Rajput
Exhibit A
Put a claim about your video archive on trial: one agent prosecutes with search, another cross-examines every exhibit against physical evidence, a judge rules, and the tape itself takes the stand. Nothing counts until it survives.
VAST DataSpaceXAI (Cursor)NVIDIACoreWeave (Weights & Biases)
View project Andrew Y Li
Replay
Replay helps find violations across video, no matter how small
VAST DataSpaceXAI (Cursor)CoreWeave (Weights & Biases)NVIDIA
View project Zain Yousaf
Receipts
The real-time video agent that remembers what happened
VAST DataSpaceXAI (Cursor)CoreWeave (Weights & Biases)NVIDIA
View project Aishwarya Patil
ghost-nyc
Safety for cyclist
VAST DataSpaceXAI (Cursor)NVIDIACoreWeave (Weights & Biases)
View project Viktor Radev
PharOS
PharOS turns NYC’s traffic cameras into an early-warning system for city safety teams, spotting near misses and risky behavior at intersections so dangerous corners can be fixed before the next crash.
VAST DataSpaceXAI (Cursor)CoreWeave (Weights & Biases)NVIDIA
View project 
mahmoudezat0121
Shepherd
A watchdog for livestock
VAST DataSpaceXAI (Cursor)CoreWeave (Weights & Biases)NVIDIA
View project 
Wanning He
FightLens
FightLens helps boxing and MMA viewers understand fast exchanges through AI-powered action tracking, contextual explanations, and evolving win-probability estimates in one viewing interface.
VAST DataNVIDIASpaceXAI (Cursor)
View project 
Subhradeep Acharjee
watcher
A multi-agent AI copilot that locks onto a vehicle or person in one traffic camera and hands it off, camera to camera, in real time.
NVIDIACoreWeave (Weights & Biases)SpaceXAI (Cursor)
View project 
Stephen MO
CarWitness 车证
CarWitness (车证) is "Ctrl+F for street cameras": search vehicle events across cameras in English or Chinese and get cited, bilingual incident reports.
VAST DataWeights & Biases WeaveSpaceXAI (Cursor)CoreWeave (Weights & Biases)
View project 
Leon
Spatial Video Agent
Spatial Video Agent lets safety and incident investigators ask one plain question across every camera and get the exact moments back, with distances in meters where the camera geometry supports it
VAST DataSpaceXAI (Cursor)CoreWeave (Weights & Biases)NVIDIA
View project Muskan Dhingra
StreetCast
StreetCast turns New York street video into cited, ready-to-file 311 reports. A rider's footage becomes every blocked bike lane, trash pile and obstruction, with the clip and the truck's fleet number, never a plate. A newsroom picks a block and gets a narrated segment where every line cites a clip.
VAST DataSpaceXAI (Cursor)CoreWeave (Weights & Biases)NVIDIA
View project 
madhu.garudala
Safe Route
SafeRoute uses AI to analyze archived street-camera footage and recommend routes around simulated hazards in an interactive evacuation drill
VAST DataSpaceXAI (Cursor)CoreWeave (Weights & Biases)NVIDIA
View project 
Vontarius Falls
World Compiler
An application that uses a relay from smart glass such as Meta Raybans to capture real time video, label it, build a semantic searchable map + digital twin.
VAST DataSpaceXAI (Cursor)CoreWeave (Weights & Biases)NVIDIA
View project 
Johaan Kathilankal Jis
Witness — AI-Powered Video Evidence Intelligence
Witness helps accident investigators and claims reviewers evaluate conflicting statements against timestamped video evidence using VAST DataEngine, NVIDIA Cosmos, and YOLO11, with human oversight.
VAST DataSpaceXAI (Cursor)NVIDIA
View project 
Michael
CloseCall
CloseCall turns the street video cities already have into measured, clip-linked evidence of near misses (close passes, cars in crosswalks, traffic squeezing work-zone flaggers) so Vision Zero and transportation teams can fix the right streets and prove the fixes worked.
VAST DataSpaceXAI (Cursor)CoreWeave (Weights & Biases)NVIDIA
View project 
Tinkamanyire “Kama” Rwezahura
Occlusive Logic
Occlusive Logic transforms video object detections into user-programmable spatial rules, enabling anyone to monitor defined areas and generate evidence-backed alerts without retraining AI models.
VAST DataSpaceXAI (Cursor)NVIDIA
View project Tanushree Umathe
Drive Recap
A trip report card for your drive; awareness score, a clip for each moment, and search; for a teen driver’s parent or anyone who wants to improve.
VAST DataSpaceXAI (Cursor)CoreWeave (Weights & Biases)NVIDIA
View project 
Chloe Liang
SecondCall
SecondCall turns the sports videos sitting in your camera roll into an instant, judge-blind second opinion: it hears and sees the decisive moment, checks it against the rulebook, and shows players and referees the evidence behind each call, starting with kendo.
VAST DataSpaceXAI (Cursor)CoreWeave (Weights & Biases)NVIDIA
View project 
freeslugs
Witness desk
Witness desk searches every camera in New York or San Francisco for one vehicle by color, body type, and time, then lets you mark the clip and keep a still.
VAST DataSpaceXAI (Cursor)CoreWeave (Weights & Biases)NVIDIA
View project 
Frank Yu
The Street Talks Back
We take street camera images of vehicles and see how compatible they are as a dating pair (tinder for vehicles)
VAST DataSpaceXAI (Cursor)CoreWeave (Weights & Biases)NVIDIA
View project