100 people in a room.
100 new company hires.
100 news feed articles.
50 potential partners.
10 remaining seats.
20 available hotels.
50 qualified candidates.
Whom should I meet?
Whom should I get to know?
Which articles should I read?
Which partner should I choose?
Which seat should I reserve?
Which hotel should I book?
Whom should I interview?
Conga's personalized social relevance
platform acts as a trusted facilitator
between people who have never met,
and offers a means to find new, relevant
people, products, services, or content.
Conga “predicts the click” between
people or groups of people.
Conga models existing social con-nections
to predict future ones.
Conga is a cloud-based platform.
Conga's patent-pending machine learning algorithms
identify patterns in an individual’s social connections
using pre-existing social data.
Once identified, Conga uses these patterns to accurately
predict new, highly relevant social connections.
Conga's science goes far beyond simple techniques like
comparing shared connections or shared interests.
Conga learns what makes an individual’s social
connections meaningful to him or her, and uses that
personalized information to predict future connections.
As a cloud-based platform, Conga makes it possible to
add social relevance capabilities to nearly any application,
across numerous use cases in multiple problem domains.
Conga is built on
Our advanced mathematical models operate on millions
of data to understand the social behavior of people,
producing accurate and scalable predictions of social fit.
should know vs
Making predictions of people
you should know is a notori-ously
hard problem, and poorly
solved by existing solutions.
Conga has created a novel ap-proach
to modeling individuals
and their relationships using
existing social data, making this
difficult problem tractable.
By contrast, social networks try
to replicate existing, offline rela-tionships.
To improve collection
of this data, basic clustering and
can be used to predict existing
that do nothing to pre-dict
The growth of social networks has been from collecting
our pasts—the people we already know. But new growth,
insights, and experiences will come from our futures,
from the people we should know.
We affectionately named our
algorithm the "ELF".
The core of Conga’s science is
the ELF algorithm, a multi-stage
Bayesian mixture model.
We named our algorithm based on its three
• Entropic, referring to our search for higher-order
• Logistic, referring to the need to model the
probabilistic occurrence of binary data such
as social connections or statements; and
• Factorial, referring to our search for
statistically independent motives.
In the case of social relationships, there are
numerous motives, or factors, hidden in the
social and profile data available from social
networks. With the right algorithm, we can infer
these factors from the co-occurrence patterns of
statements in dyadic and polyadic data.
ELF analyzes this social and profile data to
produce a set of factors that describe the social
relationship behavior of people. Importantly, ELF
goes beyond simple correlational structure to
look at the higher order statistical relationships
between statements. It models not simply
individuals or keywords, but vast dictionaries
of latent relationship factors learned from the
patterns of dyadic and polyadic relationships
found in our database.
For any given pair or group of people, ELF can
infer a small number of latent factors representing
the hidden causes which explain their relationship
or predict how a new relationship might form.
ELF analyzes social and
profile data to produce a set
of factors that describe the
social relationship behavior
Thus, Conga can identify the factors that
drive existing social connections, predict how
new, desirable connections might form, and
recommend specific factors to anchor new
relationships (e.g., sales leads).
The mathematics behind the ELF algorithm
provide unique insights into the “causes” of
relationships as mined from social data; however,
the ingestion and processing of enormous
quantities of social data could easily outpace
infrastructure without further advances.
Therefore, we have structured ELF and it's data
representation as a tree hierarchy of arbitrary
branching and depth. Queries and computation
within the system can be performed efficiently
by operating independently on any sub-branch
of the tree. The tree structure can be updated
continuously offline, and high-value branches
can be quickly identified and processed for any
realtime prediction query.
As a brief example, if we had height
and weight data for randomly se-lected
people and nothing else, then
we could define their sex, age, etc. as
These hidden “motives” for being a
particular height and weight are not
directly present in the data, but we
can infer them from the co-occur-rence
patterns of height and weight.
The ELF algorithm does the equiva-lent
for thousands of hidden factors
that predict social relationship pat-terns.
as a service.
The Conga platform runs in a public
or private cloud environment as
a service delivering social
relevance results to your
Social relevance comprises both a personalized
measurement of social fit—the Conga Score—and
insights into predictive relationship factors.
Your application retrieves social relevance results
from Conga via its RESTful Web API, making inte-gration
with a wide variety of languages, libraries,
applications, and platforms straightforward and
Conga ingests the data it needs to generate
personalized social relevance results through
either authorized access to external data sources
(like LinkedIn, Facebook, Twitter, Foursquare, and
Meetup.com), or through direct push of data from
private sources through its API.
Results are delivered on-demand, in real time, via
calls to the API. Available APIs fall into the follow-ing
• Data ingestion
• Search lookup
• Social relevance scoring insights
• Profile curation
• User messaging
• User groups
• User management
• Platform management
The heart of the Conga platform is
our personalized social relevance
The Conga platform ingests and processes
millions of semantic facts for each user to learn
on the order of 100,000 factors that collectively
describe their social relationships. The platform
then uses these factors to accurately and quickly
predict social relevance with new, unknown
people and groups.
The platform uses our proprietary machine-learning
algorithm, dubbed ELF (Entropic-
Logistic-Factorial), that is specifically designed
to work with our large-scale semantic storage
A big data infrastructure.
As a data-centric company, we use the same
petabyte-scale technologies and techniques that
industry leading companies like Facebook, Twitter,
Yahoo, Netflix, and Google use to store, process,
and analyze enormous volumes of data.
Conga ingests passive data from various external
sources, where it is normalized, canonicalized,
and then stored in a semantic form that is readily
useful for both simple queries as well as complex
For each user profile that Conga ingests, we
discover over 300 semantic facts derived from
analysis of structured, semi-structured, and
unstructured profile and social data. These facts
are then used to both generate results in realtime,
as well as feed our machine-learning algorithms
and analysis batch jobs in an offline mode.
Conga’s platform and algorithms are designed to
store and process near-limitless amounts of data,
in both transactional and batch modes. Adding
additional data does not change the fundamental
performance characteristics of the platform,
and no fundamental architectural changes are
necessary to support hundreds of terabytes of
How it works
Transactional Services Batch Services
About our technology
The Conga platform is designed to scale horizontally and
to ingest, store process near-limitless amounts of data.
Born a platform.
Conceived and built as a Web platform. Conga
is 100% Java-based and makes extensive use of
Amazon Web Services and petabyte-scale data
technologies and techniques, including Apache
Cassandra and Hadoop.
The Conga platform is accessible to clients via
industry standard RESTful (HTTP) endpoints for
easy integration with a wide variety of languages,
libraries, applications, and architectures.
In addition, these endpoints can easily be called
stubs, and endpoint metadata is available in self-describing
WADL (Web Application Description
Language) for easy tooling.
Some additional technologies we're using include
Amazon Web Services, Java EE, JPA, JAX-RS, Jersey,
Hibernate, Hibernate Spatial, Quartz, Apache,
Glassfish, jQuery, MySQL, YUI, and EhCache.
A platform that scales.
The Conga platform was designed from the
ground up to support Web-scale operation using
our experience building and deploying world-class
Web and cloud applications at NetDynam-ics
(the company that invented the application
server), Netscape, iPlanet, and Sun Microsystems.
Conga’s key subsystems are cluster-aware and
use battle-tested techniques and technologies for
• Linearly scalable NoSQL data storage
• Robust load balancing across multiple tiers
• Leveraging HTTP design principles for
efficient caching data expiration
• Cluster-aware batch processing
• Persistent queuing with cluster-aware
• Cluster-aware caching layers
• Fully stateless application request processing
Conga delivers personalized social
relevance into your products, without the
effort, complication, and distraction of
doing it yourself.
Using Conga in your products
The Conga platform has four primary capabilities that
make it useful to your product and business:
One-to-one (dyadic) social relevance scoring and insights
Given an individual, Conga returns a score (dubbed the Conga Score) representing how
likely another person is to be a good fit within the individual's social connections. In addi-tion
to the score, Conga also returns the dominant predictive factors driving the score, and
the facts that the two have in common.
One-to-many (polyadic) social relevance scoring and insights
Given an individual, Conga returns a score representing how well that individual fits within
a group of people. The higher this score, the more likely the individual is to be a good fit
within the group. In addition to the score, Conga also returns the dominant predictive fac-tors
driving the score, and the facts that the polyad have in common.
Ranking and sorting of groups of people
The Conga platform can rank and sort groups of people according to social relevance to a
given individual. Groups can be hyper-segmented across tens of millions of unique facts,
including geotemporal location, as well as ad-hoc, curated collections.
When combined, the capabilities above can be used to find and surface the people—and
by proxy, content and products—that should be recommended to an individual.
Conga helps add meaningful
Taking products in domains that aren’t traditional-ly
social, like e-commerce, and turning them social
is an opportunity to differentiate and delight.
By providing social relvance, Conga helps trans-form
your product into a meaningful, relevant
social experience, even when your customers
don't already know each other.
Conga differentiates your app
from the pack.
Personalization is the secret sauce you can use to
make your product stand out from similar offer-ings,
and it gives you a new means for attracting
loyal, delighted customers.
Conga helps personalize your product by provid-ing
social context tailored to your customer.
Conga provides relevant, new
avenues for discovery.
Conga helps uncover people in your customer
base that are similar to a customer’s social connec-tions.
These people then act as proxies to help your
product make interesting and relevant recom-mendations—
as relevant and useful as if coming
from an actual connection.
Conga adds social relvance to your
Adding a layer of personalized social relevance
to your content, products, and services can help
elevate what may otherwise be a mundane cus-tomer
experience to something new and fresh.
Conga's social relevance can directly power peo-ple
discovery in your product, as well as provide a
proxy for content and product discovery.
Conga unlocks innovative, premi-um
opportunities for your users.
Where available, airline customers have shown an
openness to pay for the ability to choose a seat
based on interpersonal click with fellow travel-ers,
a capability called social seating.
Conga can help power similar opportunities
across the hospitality, service, and travel verticals.
Conga is a turn-key, cloud-based
Don't spend your valuable time building a com-plex
feature that isn't your core value proposition.
Get your app off the ground faster and ensure suc-cess
by integrating Conga’s technology into your
product rather than trying to build it yourself.
Market: Conferences Events
Increase the value of events by automating discovery of
social connections between attendees.
Attendee satisfaction and repeat attendance depend on attendees get-ting
the most out of your event.
Organizers Add a new layer of social relevance to your conferences and events by automati-cally
Quickly identify relevant social connections between attendees, speakers, ex-hibitors,
or sponsor, and tap into new opportunities for networking.
tailoring recommendations for each attendee and providing new, relevant
avenues for interaction between event participants.
» Make your event stand out through more satisfying, personalized attendee experiences.
» Open new, meaningful channels for social interaction.
» Tailor content recommendations for each attendee
» Discover who among attendees are best matched.
» Increase attendee-to-attendee collaboration
» Improve the attendee experience
» Make your events more valuable
With Conga, your attendees will get more
value out of your event by accessing
personalized content and networking
opportunities that improve attendee
collaboration and customer satisfaction.
Find out how a candidate will fit with your team—before
the first interview.
Team dynamics are a key driver of any successful organization, but when
building a team it can be hard to quantify the softer side of candidates, to
obtain a complete picture of how they fit with existing team members.
Recruiters Conga reveals underlying factors to measure how a candidate fits with individ-ual
team members as well with the larger team, giving valuable insight before
you make the hiring decision.
» Improve team efficiency and morale by putting together the most compatible people
» Measure how candidates will fit with your team
» Get valuable social insight before you hire
» Filter and rank candidates by team fit
» Find similar candidates
Conga gives you valuable insight into a
candidate's social fit with your team and
helps you identify the right candidates
faster, saving you precious time during the
Market: Travel Hospitality
Seize a new opportunity to modernize and personalize
travel hospitality experiences.
Help travelers choose flights, seats, and hotels based on social relevance
with other travelers to increase customer engagement, brand affinity, and
Airlines Personalize seating recommendations for each passenger and suggest the ideal
Get the guests you want. Let them know that “guests like them” stay with you.
Give guests a chance to discover each other.
seatmate based on a prediction of social fit between passengers.
» Make flight and room selection about people, not just price
» Personalize seating recommendations
» Engage guests with your brand—your venue
» Delight customers drive loyalty
» Charge a premium
With Conga, your customers will be able
to choose their airline and hotel based on
social fit with other travelers, increasing
customer engagement and maximizing
Market: Social Business
Help employees connect more efficiently and get work
Even in the smallest organizations, it can be difficult for employees to
identify relevant connections with other employees, creating organiza-tional
silos that hinder efficiency.
Employers Recommend relevant people and groups to follow and join, and tailor content
recommendations for individual employees. Personalize content for each em-ployee
so they can quickly discover peers outside of their normal work groups
and access expertise they need to get work done faster.
» Give your employees the social insights to connect effectively
» Create enriching mentor-mentee programs
» New employee introduction indoctrination programs
» Insights on the professionals they need to know
Conga helps companies break down
organizational silos by automating the
discovery of relevant social connections
between employees, enabling faster access
to information expertise, and increasing
For detailed information on how Conga
can help your business, address specific
product requirements, or be applied
to a particular market, please refer to
additional information included with
this presentation, or contact us directly.
Born in Washington D.C. Lived in MD, VA, NC, PA,
WI, and MA before settling in California.
Built ham radio receivers and programmed bio-rhythm
charts in machine language as a kid.
Surprising range of industry experiences includ-ing
nuclear reactor control systems, paper making,
water treatment technology, voice over IP service,
filtration fabric, DSL, coated and printed paper and
film, and Wi-Fi service.
Entrepreneur (founder to $100M+ exit), senior ex-ecutive,
board member, advisor, investor, salesman,
product manager, engineering manager, business
development executive. Investor, board member,
advisor, mentor, conference presenter, panelist,
Three patents issued or pending.
BS Electrical Engineering, Duke University
MS Nuclear Engineering, Bettis Reactor Eng.
MBA Fuqua School of Business, Duke University
Family man. Competitive masters swimmer.
Raised and schooled in Texas. Landed in Silicon
Valley by way of Austin and Denver and intends to
Wrote games in 6502 assembly in 8th grade. Later,
wrote the music for a famous Hollywood director's
Drummer, composer, designer, photographer,
writer, public speaker, philosopher. Had a fling with
rocket science and steam, but returned to software
and lived happily ever after.
Architect and coder with a penchant for product
design and user experience. Has written a few app
Chief Architect for Sun’s Platform-as-a-Service and
Java Enterprise Tools groups founded Zembly.
com, a cloud-based IDE and social development
Four patents issued or pending.
BS Mechanical Engineering at UT Austin
Is the human Shazam for Seinfeld.
A valley-man, born and bred in Silicon Valley. San
Franciscan, ice hockey player, reformed commuter.
Product Manager at Zembly.com (Sun Microsys-tems),
a cloud-based IDE hosting platform that
enabled browser-based development of apps for
Facebook, iPhone, and social platforms.
Product management, marketing, engineering at
Sun Microsystems and Borland Software.
Has worked in developer tools, cloud platforms,
location-based services, social discovery apps.
BS Operations Management Information
Systems, Santa Clara Leavey School of Business
MS Software Engineering, SJSU
Ask him about baseball.
Likes to predict. Can read your writing and tell you
with 97% confidence what you’ll be doing a year
from now. Scary smart yet extraordinarily gracious
to the rest of us.
Entrepreneur. Founded two companies, sold one,
and operates another in her spare time.
Uses expressions like “Efficient/Sparse Coding,”
“LDA and topic modeling,” and “Multi-modal data
mining” with the same ease the rest of us use
“grande non-fat latte.”
BS Cognitive Science Neuroscience, UCSD
MS Psychology, Cognitive Neuroscience, CMU
PhD Psychology Theoretical Neuroscience,
Visiting Scholar, UC Berkeley as Research scien-tist
at the Redwood Center for Theoretical Neu-roscience
• Conga, Inc. was incorporated February 2010 as a
• 100% of shares authorized or outstanding are
• Conga is privately funded.
549 Laswell Ave
San Jose, CA 95128