[a little blurry picture of
Paul M. Aoki]
Paul M. Aoki
Independent consultant
Machine learning student

Overview Publications Service Personal

I work in the areas of technology and international development, human-computer interaction and computer systems. I'm a research scientist by trade, currently on a self-funded sabbatical to study the mathematics underlying statistical learning.

Briefly, the facts of my life as are follows:

SSWE Google Google 2011-2016
SSRS Intel Research Berkeley Intel 2006-2011
MRS II Computer Science Laboratory PARC 1999-2005
Ph.D. Computer Science U.C. Berkeley 1993-1999
Lt. Surface Warfare U.S. Navy 1989-1992
B.S. Electrical Engineering & Computer Sciences U.C. Berkeley 1984-1988

Research Interests:

Technology and International Development

As a software engineer at Google, I worked with the Access Strategy & Operations team (originally part of Google.org) on sustainable solutions to access problems in developing regions. Examples of the team's projects (in collaboration with many other teams at Google) included pilots of mobile data tools (SmartBrowse) and TVWS (in South Africa). The StratOps team developed plans for network deployments (access strategy) and helped execute those that move forward (operations). These ranged from the fairly conventional (metro fiber backhaul in Uganda and Ghana, urban Wi-Fi networks in India, satellite-fed CDN in Southeast Asia) to the extremely risky. We had to model everything from orbital patterns, wireless link budgets and network traffic up to local consumer economics.

As a research scientist at Intel, I worked on TIER, a project in which Intel collaborated with U.C. Berkeley on the design of appropriate technology for emerging regions. Various TIER sub-projects had connections with Intel's former Emerging Markets Platforms Group (EMPG). My published research was based on fieldwork in Ghana and Kenya and a field visit to Chile.

Human-Computer Interaction and Computer-Supported Cooperative Work

My work in HCI has been highly interdisciplinary, mostly sitting in the intersection of mobile computing and human interaction.

I still participate in a project on mobile participatory sensing called Common Sense. The team at Intel Research explored some technical issues in sensing environmental air quality using mobile phones, but the key problem of interest for us as HCI researchers has been to work with partners such as the West Oakland Environmental Indicators Project (WOEIP) to understand how and why people become engaged in citizen science and environmental activism. (Note that environmental degradation is a key concern in emerging regions.)

I collaborated for many years with PARC researchers in the sociological discipline of conversation analysis; their detailed understanding of human conversational structure have helped us understand how (technologically) minor differences in a system can have dramatic effects on interaction.
...An example of this work is a system we nicknamed Mad Hatter, a mobile audio space designed for use by small gelled social groups. This work applied machine learning techniques to produce a conferencing system that adapts to human conversation.
... As part of this work, Allison Woodruff and I analyzed some fieldwork on young adults using push-to-talk cellular radios, drawing on several perspectives in computer-mediated communication.
...Another example is Sotto Voce, a networked electronic guidebook system that delivers audio information in a way that facilitates face-to-face human interaction instead of inhibiting it.

All of this work on interaction has also lead to a renewed interest in wide-area collaborative systems. I spent a lot of my time at sea sitting in front of networked tactical command-and-control consoles.

Finally, I have some background in visualization: information display technology, information visualization techniques and visualization system architectures. This came through working with members of the Tioga DataSplash database visualization project at Berkeley.

Data Management

As part of TIER, I had the experience of architecting, building and deploying an asynchronous data replication system for use in a telemedicine system in Ghana (2007-2008).

I consulted (2002-2003) for the PARC security group in a project on privacy preservation in distributed repositories of personal data.

The Harland project at PARC (1999-2000) explored the programming affordances of data modelling constructs that lie between the semistructured and (dynamic) object-oriented data models. Another consideration is how such constructs can be supported efficiently. Harland was an outgrowth of the PARC Placeless Documents project on property-based data stores.

At Berkeley, I worked for Mike Stonebraker in the database research group on a variety of topics: [link to UCB
database research group]

Paul M. Aoki, aoki@acm.org
Modified: $Date: 2017/01/12 01:24:52 $ by $Author: pmaoki $