Robert Munro
/ Rob Munro

Robert Munro is an expert in combining Human and Machine Intelligence, working with Machine Learning approaches to Text, Speech, Image and Video Processing. Robert has founded several AI companies, building some of the top teams in Artificial Intelligence. He has worked in many diverse environments, from Sierra Leone, Haiti and the Amazon, to London, Sydney and Silicon Valley, in organizations ranging from startups to the United Nations. He most recently ran Product for AWS's first Natural Language Processing services in the Deep Learning team at Amazon AI.

Robert has published more than 50 papers and is a regular speaker about technology in an increasingly connected world. He has a PhD from Stanford University.

When not working Robert travels the hidden corners of the world by bicycle, most recently cycling across Alaska. He shares his thoughts at @WWRob, Jungle Light Speed.

His previous work, in his words:

Cloud-based Natural Language Processing

I lead AWS's first Natural Language Processing products from conception to internal launch. It was incredible experience to learn Amazon's product development process, while leading the first suite of products in my area of expertise - an experience that not many people get. Before that, I co-founded and was CEO of Idibon, a ~40 person AI startup in San Francsico. I drove our vision, decisions and strategic direction. It was a pleasure leading one of the top teams in AI towards the shared vision of bringing language technologies to all the world's languages.

AI and Distributed Human Computing for Disaster Response

I am the recognized global expert in Crowdsourcing and Natural Language Processing for disaster response, having helped crisis-affected populations and disaster response professionals respond to dozens of large disasters, both human and man-made, in more than a dozen countries, from floods in Pakistan to violence in the Middle East and hurricanes in the USA.

I introduced crowdsourcing to the disaster response community when I coordinated the translation, geolocation and categorization of emergency text messages sent in Haiti in the wake of January 12, 2010 earthquake. I worked with 1000s Kreyol and French-speaking volunteers from 49 countries, turning text messages in Haitian Kreyol into categorized English messages with precise coordinates. For 80,000 messages, the typical turnaround was just 5 minutes: their vital local knowledge saved hundreds of lives and directed the first aid to tens of thousands.

Similarly, with Natural Language Processing I showed how machine-learning can help identify and prioritize urgent reports and track trends in the quickly changing post-disaster environment. This ultimately formed part of my PhD at Stanford.

My experiences have left me profoundly impressed with how crisis-affected communities are able to respond to tragic and often sudden events by bootstrapping their own recovery. My experiences have also left me believing that the not-for-profit community does more harm than good following disasters, especially in information management, with the international success coming from professional technology companies whose products are tried and tested more broadly. This was one central motivation to create Idibon and we are proud to support disaster response both locally and globally.

Global Health Monitoring

I have maintained a strong interest in technology for tracking global health, especially in monitoring and preventing epidemics.

In 2011 I worked at Global Viral Forecasting (now Metabiota) as the Chief Technology Officer for EpidemicIQ, a system for tracking disease outbreaks world-wide. The goal is to predict and prevent future epidemics. This was ground-breaking work with a skilled and diverse team, enabling us to track disease outbreaks from billions of data-points daily from data in more than a dozen languages, often beating the major health organizations by days in identifying outbreaks.

Like my disaster response work, it also relied heavily on machine-learning and crowdsourcing. One innovation was the use of online games to pay workers. When there was an E-Coli outbreak in Germany, we helped track the outbreak by paying people virtual currency to help process the reports. In one case, the currency took the form of virtual 'seeds' in a farming game: so by playing an agricultural game online, German-speakers were helping track a real agricultural outbreak outside their doors.

This work was presented to the UN General Assembly on Big Data and Global Development, among other places. See my post in tracking epidemics for more info.

Clean Energy

Prior to moving to the US I lived in Sierra Leone, working for the Environmental Foundation for Africa and for the United Nations High Commission for Refugees in Liberia, installing solar power systems in schools, clinics and national parks. I joined Energy for Opportunity as the Chief Information Officer at the organization's conception and managed the information and communication systems for our West African-based Solar Power operations, helping the company from its launch to see it grow to become the leading renewable energy provider in the region.

It has been a joy to see Energy for Opportunity become one of the largest power companies in Sierra Leone, a place where the cheapest energy is also often the greenest. I stepped down as CIO of Energy for Opportunity when I became CEO of Idibon, but I remain active in the company and in the Bay Area energy community.

Understanding the Diversity of Languages

Providing resources in a person's first language allows them to negotiate the connected world on their own terms.

As much as half of the world's 7,000 languages will disappear within the next century. The majority are only spoken languages, so the race is on to record as many as possible before all trace of the language, stories and culture are lost. As part of the same globalizing outcomes, we now have the technologies to record and study the world's languages for the first time, meaning we have a thin sliver of time to capture much of humanity that will otherwise be lost.

I have worked on architectures and strategies for recording the world's languages for more than a decade, including early work on the The Hans Rausing Endangered Languages Project, which supports the (often unique) recordings of 100s of languages. In July 2011, I also helped run the Workshop on Crowdsourcing Technologies for Language and Cognition Studies, bringing together the researchers who are embracing the ability to interact with people and study their communications via new technologies and experimental methods.

More directly, I also had the privilege to live with the Matses of the Amazon and study and record their language, one of the most unique on earth.