working on making cities ‘Smarter’ for over a decade

projects.

implemented AI based mission critical Incident management solutions to reduce the incident response time.

Jun 2016 - Jul 2019

FUSION ANALYTICS FOR PUBLIC TRANSPORT EMERGENCY – SINGAPORE LTA

I directed and executed the Fusion Analytics for Public Transport Emergency (FASTER) project for the improvement of Singapore’s MRT services. The goal for this project is to provide optimized response plans in case of transit incidents. I designed a system which implemented complex fusion analytics based on WIFI, transit fare card and video data to identify incidents, predict the impact of incidents and generated multiple response plans using optimization algorithms. The fusion algorithms utilized WIFI and ATS data to determine Train and commuter KPIs for the overall transit network. I led a team of architects and developers based in India, Korea and Singapore to iteratively design, develop and deploy the models generated by data scientist team.

Jan 2015 - May 2016

TRAFFIC DECISION CONTROL - DECON - SINGAPORE LTA

I designed and implemented an advanced traffic incident detection and response system that featured state of the art streaming architecture, AI and IoT technologies. Based on real time situations and predictions, DECON optimizes traffic flow and safety by adjusting traffic signal times, variable message signs (VMS) and variable speed limits.

Jan 2013 - Dec 2015

NJTA INTELLIGENT TRAFFIC MANAGEMENT SYSTEM

I designed and implemented a system having an integrated view of all signs along with network view and advanced analytics tools to automate updates to the speed limits and travel messages for drivers. This engagement with NJTA leadership helped us displace the incumbent, Delcan and they signed a multi-year implementation agreement with IBM to enhance the Authority’s traffic management system. This was the first instance worldwide that IBM's Transportation analytics solution was implemented to provide a holistic view for monitoring traffic events and conditions on two of the world's most heavily traveled highways and busiest toll roads in the United States.

Aug 2012 - Feb 2014

TRANSIT ANALYTICS – IBM INTELLIGENT TRANSPORTATION

I was the Architect for the Transit Analytics component of IBM Intelligent Transportation. I developed a Service Oriented Architecture which supported interfacing with multiple bus operators, traffic systems and third party applications. I created an architecture having the following components - Arrival Prediction, Historic Calculation, Bus Information, Traffic Information and Bus Operations.

Jan 2010 - Jun 2011

BUS ARRIVAL PREDICTION – SINGAPORE LTA

I drove a 150% improvement on the accuracy of estimated bus arrival times, from +/-1 minutes error down to +/- 30 seconds. Conventionally, this would not have been possible, due to the highly variable nature of road conditions. I successfully combined Traffic Prediction models with GPS information from buses to create a dynamic system capable of processing massive variable data and producing predictions that are accurate within seconds of bus arrival.

Jun 2008 - Aug 2013

TRAFFIC PREDICTION – IBM INTELLIGENT TRANSPORTATION

I led of a team of application developers, administrator and deployment experts to design, develop and deploy Traffic Prediction at Singapore LTA and NJTA. Post the successful client implementations, I led the productization of Traffic Prediction component. In the planning phase, I worked with the Intelligent Transportation product manager to define release based requirements. I worked with the usability expert to design the administrator and operation user interface. I created an UI design which supported global languages and could also be used by visually challenged.

Feb 2009 - Nov 2011

TRAFFIC PREDICTION AND INCIDENT DETECTION – TRANSPORT FOR LONDON - TFL

Transport for London was interested in implementing an Automatic Incident detection system. Typically incidents were being reported in about 10 to 15 minutes. This delay at times resulted in heavy congestion. TFL wanted to reduce the incident detection time to no more than five minutes. I worked with IBM Research team to create the models for Incident Detections and Impact Predictions. Based on the requirements and models developed, I defined the following components - Traffic Information Hub, Traffic Prediction, Incident Detection Module, Incident Severity Evaluation, Traffic Simulator and Decision Support System Optimizer. I led the implementation team for delivering a Cloud based solution.