Center for Smart Streetscapes VALIDATE Accelerator
This program helps early-stage innovators build and scale their ideas through hands-on education, mentorship, and business model development. Offered by Columbia University in partnership with other institutions in the NSF Engineering Research Center for Smart Streetscapes (CS3) and Columbia’s Start Me Up Bootcamp, the program is tailored for startups in the smart cities space.
It includes two main modules:
- Product–Market Fit & Customer Discovery: Learn the Lean Launchpad methodology, conduct field interviews, and validate your business model.
- Smart Cities + Urban Tech Education & Development: Gain targeted insights into technologies, challenges, and opportunities in the smart cities sector.
Our fourth cohort application opens August 3. Students, faculty, and researchers at Columbia, FAU, Lehman, and Rutgers are eligible to apply. If you're considering applying—or know someone who might be a great fit—complete the interest form below to receive an application reminder and program updates.
Fall '26 Accelerator Cohort Coming Soon!
Our Alumni Teams
Powered by
EDGE AI FOR STREETSCAPES
Glyph | Cliff Gelb, Columbia Business School
Glyph uses edge AI to sense real-time streetscape signals and instantly create on-brand promotions for independent restaurants in dense cities. These ads deploy to digital signs in front of and within the restaurant, helping them smooth demand, fill tables, and keep neighborhoods vibrant.
AeroBin | Azra Bano and Yusuf Lee, Rutgers University-New Brunswick
AeroBin is a smart city platform that uses edge AI sensors to monitor waste fill levels and air quality in real time. By predicting overflow and pollution spikes, AeroBin helps cities optimize waste collection, reduce emissions, and improve urban livability through a digital twin-driven insights dashboard.
SAFE AND PRIVACY-PRESERVING SENSING
KIDOVA | Jiayi (Penny) Dai and Amily Li, Columbia Graduate School of Architecture, Planning and Preservation
Kidova is a child-friendly platform improving children's movement and feelings through the city. By guiding safer school routes, fostering independence, and sharing insights with digital models of streets, it helps families feel secure while supporting communities and planners in building more walkable, child-friendly neighborhoods.
DIGITAL TWIN PLATFORMS
UrbanMirror AI | Jonathan Lalla, Florida Atlantic University
UrbanMirror AI is building a SaaS digital twin platform that enables cities, universities, and developers to simulate streetscapes, traffic flow, and policy changes in a virtual environment. By integrating IoT data and AI, our solution reduces planning risks, improves safety, and supports smarter, more sustainable urban development.
NaviMind | Mengxuan Liu, Carlos Espinoza, and Yongjie Fu, Columbia Engineering
NaviMind is an AI navigation assistant powered by crowdsourced vision data across public transit settings. With VR/AR glasses, commuters capture images that enrich the platform and, in return, receive real-time AR overlays for navigation, accessibility, and safety. NaviMind co-curates mobility experiences that are seamless, resilient, and community-driven.
Arachna Networks
Team: Jacob Kahn, Gabriel Garcia
Advisor: Dr. Georgios Sklivanitis
Affiliation: Florida Atlantic University, Undergraduates in Computer Engineering with faculty advisor
Arachna Networks wants to implement emergency fast deployable networks to smart city enchaining utilities that operate with a full independent network stack that is not centralized. We aim to bring Arachna Networks to public safety, law enforcement, and defense markets. Arachna Networks is an open-source based solution built on commercial off-the-shelf low-cost hardware. During the program, our goal is to validate if Arachna Networks will be a software-hardware or software-only product and develop a commercialization plan of our solution.
InfraInsight
Team: Ines Khoulder, Samarth Agrawal
Affiliation: Columbia University, Undergraduates in School of Engineering and Applied Science
Building Managers face challenges in optimizing and tracking energy efficiency given the high cost of IoT retrofits. Although scaling is hindered by the wide variety of building types, augmenting “digital-twin” building solutions with satellite and remote sensor data can provide good-enough quantitative insights to affordably model potential improvements. This project is the confluence of 2 projects undertaken throughout Fall 2024 as part of academic coursework. The team hopes to pinpoint the ideal area of product-market fit and come to a concrete go-to-market strategy.
S.T.R.I.D.E
Team: Muhammad Shahbaz, Md Mahmudul Islam
Advisor: Dr. Shaurya Agarwal
Affiliation: University of Central Florida, PhD student and Graduate Research Assistant in Civil, Environmental, and Construction Engineering with faculty advisor
Urban Sentinel aims to improve traffic sensing and management using an AI-enabled software suite that seamlessly integrates multiple sensors -- cameras, lidars, radars and more, regardless of manufacturer, to accurately detect, track, predict, and analyze road-user trajectories. Using multi-modal deep learning models for advanced sensor fusion and large language models fine-tuned for reasoning, Urban Sentinel empowers cities by providing decision-level intelligence. The application areas include but are not limited to traffic management, traffic safety and accident prevention, and support for connected and autonomous vehicles.
Viarithm
Team: Dr. Mehmet Turkcan
Affiliation: Columbia University, Associate Research Scientist in Civil Engineering and Engineering Mechanics
Viarithm develops and openly shares reliable computer vision models for urban environments, backed by comprehensive real-world datasets and evaluation frameworks. They enable trustworthy AI applications in cities by providing thoroughly tested models, hardware and transparent evaluation metrics that reflect actual street conditions. Viarithm is the only platform providing the complete stack: datasets, models, evaluation, and hardware deployment.
SatSight
Team: Jackson Ye, Joseph Bajor
Advisor: Dr. Alexandre Morozov
Affiliation: Rutgers University, PhD student in Electrical and Computer Engineering with faculty advisor
SatSight is an affordable AI platform that analyzes satellite imagery in real time, providing urban planners with up-to-date insights on traffic flows, infrastructure conditions, and environmental health, enabling proactive, cost-effective decisions that promote sustainable growth, reduce pollution, and enhance safety. The technology stands apart from competitors due to its cutting-edge Object-Centric State Space Models (SSM) AI model, which achieves near-linear memory complexity while delivering a global perspective on urban dynamics.
SINA
Team: Bowen Fang, Ruijian Zha, Hongcheng Tian
Advisor: Dr. Sharon Di
Affiliation: Columbia University, PhD and Master’s students in Industrial Engineering and Operations Research with faculty advisor
SINA differentiates itself by offering context-aware and highly personalized route recommendations that go beyond what competitors like Google Maps or Citymapper provide. The app incorporates real-time contextual information derived from natural language inputs, images, and integrations with third-party applications such as Notion. By dynamically syncing with calendars and other productivity tools, SINA can proactively adjust plans based on real-time data, such as unexpected delays, weather changes, or new user priorities. SINA processes diverse and complex user inputs, such as dash cam images or detailed queries, to provide actionable insights that competitors cannot match. By combining advanced AI capabilities, real-time adaptability, and third-party integrations, the solution redefines transit planning for a smarter, more responsive user experience.
Team: Marco Tedesco
Affiliation: Columbia University, Research Professor at the Lamont-Doherty Earth Observatory
LinkedIn: Marco Tedesco
Email: [email protected]
Climate Impact detects flooded city areas using web cameras and images as well as artificial intelligence technology, providing estimations of water levels in impacted areas. Climate Impact’s detector has both stand-alone and plug-and-play capabilities, allowing it to generate ubiquitous real-time, accurate estimates of flooding. Through the Climate Impact product, cities can better monitor flood trends and mitigate the effects of flooding on their infrastructures and communities.
Team: Mahshid Dehkordi, Jeremy Johnston
Affiliation: Columbia University, Columbia PhD Candidates in Electrical Engineering
LinkedIn: Mahshid Ghasemi, Jeremy Johnston
Email: [email protected], [email protected]
Cyrus is an AI-native security system that can identify an extensive range of abnormal behaviors and activities that can endanger safety and security standards in indoor and outdoor areas. Cyrus can turn any security cameras (regardless of their manufacturing brand) into a non-intrusive safety and security patrol system that works in real-time. Cyrus does not require fine-tuning and retraining to adapt to a new environment. Cyrus can detect subtle indicators of safety concerning behaviors or objects, such as verbal or physical altercations and various types of weapons, with high accuracy in real-time without any extra fine-tuning.
Team: Caspar Lant
Affiliation: Columbia University, PhD Candidate in Computer Science
LinkedIn: Capsar Lant
Email: [email protected]
In an era of increased risk from vehicular traffic, Walkwise protects the most vulnerable road users. Through their GPS-based pedestrian intent prediction tools, which leverage a transformer-based machine-learning model that inputs live, local GPS-positioning data (as recorded by a user’s phone), Walkwise provides information on pedestrians intersection interactions. By transmitting crossing predictions to intersection signaling gateways, Walkwise can provide dynamic, optimal solutions for any traffic circumstance.
Team: Georgios Sklivanitis, Dan Zimmerman
Affiliation: Florida Atlantic University, Faculty in Electrical Engineering and Computer Science; Florida Atlantic University, Graduate Student in Electrical Engineering and Computer Science
LinkedIn: Dan Zimmerman, Georgios Skilvanitis
Email: [email protected], [email protected]
FloodSense improves the accuracy of flood forecasts in areas lacking weather infrastructure. By crowdsourcing hydrologic data, FloodSense offers an entirely new data economy that rewards people and cities who deploy and maintain FloodSense’s flood monitoring sensors. Through a dense, decentralized flood marketplace that users can purchase and sell hydrologic data on, FloodSense can improve citywide flood awareness.
Team: Christopher Grullon
Affiliation: Columbia University, Undergraduate Student at the Fu Foundation School of Engineering and Applied Science
LinkedIn: Christopher Grullon
Email: [email protected]
Sustainable Safe Smart Intersections is equipping cities with technology that enables more sustainable living. Through their intersection simulation model, Sustainable Safe Smart Intersections offers an analysis of the impact of intersection redesigns. Trained on integrated, real-world data points, the product offers simulations similar in accuracy to real-life, intersection situations.
Team: Izaac Martinez
Affiliation: Florida Atlantic University, Undergraduate Student in Data Science and Analytics
LinkedIn: Izaac Martinez
Email: [email protected]
The FloodFinder product detects road flooding, alerting the public of optimal rerouting options. The FloodFinder product leverages both FloodFinder cameras’ feeds and pre-existing traffic cameras to identify flooding. Using neural networks to develop software that analyzes photos and video, the FloodFinder alarm light will warn the public about high-flood areas, keeping drivers safer and better informed.
Program Information
- At least one team member must be affiliated with a CS3-funded project. Otherwise, applicant team must provide a letter of support from a CS3-affiliated faculty that highlights how the proposed idea would help advance and accomplish the Center’s mission.
- Applicant team must be working on an idea or prototype for an innovative smart city solution within the 3 center thrusts (below) or on an idea or prototype similar to these thrusts:
- Edge AI for Streetscapes: Edge AI for urban mobility and infrastructure monitoring
- Safe and Privacy-Preserving Sensing: AI for anomalous behavior; Privacy-by-design methods for collecting data in public spaces
- Digital Twin Platforms: Policy simulation and streetscape planning
- Team must be ready to commit 5-10 hours per week for customer discovery, business validation, and product build for their ventures
- Team has received no institutional funding at the time of application (grants and awards are acceptable)
- Preference for two team members per team – both willing to commit to program’s requirements (I-Corp and CS3-specific programming)
- Team must commit to attending ALL weekly sessions
- Applications will be reviewed by Columbia Technology Ventures and CS3 leadership team to determine fit within the field/area of interest
- Each application will be judged based on team quality, commercialization potential, and scientific and technical merit
- Columbia teams may apply for travel funding to continue customer discovery and attend a conference or industry event. Bootcamp requirements must be completed by the team including 20 customer interviews, attend bootcamp sessions and office hours. Funding support levels vary for teams $1,000 – $3,000.
- Stipends are available for bootcamp team members who complete all requirements with a financial support of $500. Underrepresented students and post-doctoral researchers are encouraged to apply.
- Upon completion of the program, teams are eligible to apply for the National Science Foundation’s I-Corps Teams program, which includes a $50,000 grant.
- This year’s winner will secure a guaranteed spot in NEC-X’s 2026 Elev X! cohort.
- Teams that have successfully completed the multi-week program and are working in areas that align with the CS3 mission, will be eligible for consideration of an award of up to $10,000 to support advancement of their projects.
Email: L2M@ctv.columbia.edu
