About iCog - Labs Jobs in Ethiopia

Company Profile
 
Technology has advanced a long way since the creation of the first computer. The notion of an intelligence explosion has become almost common topic in some circles, with pragmatic and acclaimed visionaries such as Ray Kurzweil arguing that AIs with general intelligence beyond the human level can be expected by the 2040s.
As the first private AI and Robotics research and development company in East Africa,we share this optimistic outlook. We intend to be on the cusp of this intelligence explosion, perhaps even earlier than the 2040s.
iCog Labs is a software development company based in Addis Ababa, Ethiopia focusing mainly on Artificial Intelligence research and development.

Our Vision

The work we do at iCog Labs is quite practical, making steps forward in various AI research areas and building software for our customers. But this day-by-day practical work is guided and inspired by a broader vision.
As partners of international future-oriented enterprises such as Humanity+, OpenCog and Novamente LLC, we are dedicated to a grand vision—we dream of making a breakthrough in the capability of AI systems to think, learn and create, and using this breakthrough AI to improve the lives of people in Africa and around the world.

 
Our History

iCog was co-founded in 2013 by young Ethiopian entrepreneur, Getnet Aseffa Gezaw and an American AI pioneer, Dr Ben Goertzel. Ever since its establishment, 5 years ago, iCog Labs has been working for and with international customers Generally,the work being done at iCog Labs focuses on Artificial General Intelligence, Robotics, Artificial Intelligence enhanced Commercial Software, Bio-informatics Research, Machine Learning based data analysis, and recently there has been some work being done in the area of prototype development on some Hardware products.
 
Currently it has 21 programmers and engineers working on a full time contract and more than 30 junior programmers working on par-time arrangement. The team of programmers at iCog Labs have an aspiration to advancing the utilization of AI techniques.
 
iCog carries out its works in partnership with a number of international organizations including: US AI firm Novamente LLC, US bioinformatics firm Biomind LLC, Dr. Gino Yu's lab at Hong Kong Polytechnic University and the OpenCog Foundation.

The company, for the past three years, had been working closely with two Ethiopian universities namely Addis Ababa University (AAiT campus, and 4 Killo campus of natural science) and Mekelle University. Hitherto, we had been collaborating, with the respective industrial link of these two universities and we became one of the major stakeholders in their internship programs in the fields of Electrical, and Computer engineering. In addition to these two universities, as of writing, iCog has a signed agreement for apartnership with 24 public universities in Ethiopia, one public university in Nigeria and as well with the Federal Ministry of Science and Technology. iCog pursues the idea that computers can one day simulate the entire human brain. The company also focuses on practical applications of AI for various clients around the world.

 
Products, Services and Customers

We have served our customers with a variety of projects, but our core specialty is Artificial Intelligence which includes machine learning based data analysis, computational linguistics, computer vision, mobile robots and cognitive robotics, cognitive architectures and artificial general intelligence. Our staff brings experience contributing to international AI development teams, as well as software and hardware development experience on Ethiopian government and industry projects.
 
iCog Labs is a company that mainly focuses on AI. That being said, it doesn't mean that all it does is AI R&D but also serves customers by developing custom software attuned to their specific needs. iCog Labs' current projects include Hanson Robotics (vision processing and perception software for the robots), Opencog development, Medelert 2.0 or The Virtual Health Assistance (a software utilizing the Telehealth technology), Hershey's Trend Prediction software, Pattern Recognizer for Longevity Research (customized for Stevia First PLC), Autonomous Customer support software for a company called Build Direct, Yanetu Software (a Teaching Tablet), Online Music Recommender software for Polo Grounds Music, a subsidiary of Sonny Music.

 
Hanson Robotics

Hanson Robotics is a robotics R&D company that specializes in humanoid robots. They have done an amazing amount of work in the area. They have developed robots with facial textures and expressions that have an uncanny similarity to humans. Their goal is to introduce to the world robots that have the most human like features with a full-range of facial emotion expressions and trustworthy behaviors.
 
“We hope to create a better future for humanity by infusing artificial intelligence with kindness and compassion, achieved through millions of interactions between our robots and the people whose lives they touch. Our long-term mission is to dramatically improve people’s everyday lives with affordable, highly intelligent robots that teach, serve, entertain, and are capable of developing a deep relationship with people. In time, we hope our intelligent robots will come to truly understand and care about people and evolve greater than human wisdom, to the point that they will one day be able to address and solve some of the most challenging problems we face.”
Dr. David Hanson, Founder and CEO of Hanson Robotics
 
At this time, the company has built many robots with the above specifications including Albert Einstein HUBO, Sophia, Han, Jules, Joey Chaos, Alice-Eve, Bina48, Philip K. Dick and Diego-San. Their most respected specialty is the head. The robot heads they build show facial expressions according to emotions they are supposed to feel in a certain situation like surprise, anger, happiness, sadness and many others. Since Dr Goertzel joined the company
as Chief Scientist, they have integrated the Opencog platform for the robots' intelligence in speech and decision making.
The robot heads can perform well by the emulation of over 62 facial and neck muscular architectures, vision through cameras on the eyes and neck which serve for facial and object recognition and tracking, voice identification and speech recognition for the chat system and a broad API for collaborative modifications. Not all but most of the code the company uses to drive these systems is open source and can be found on Github but developing projects and proprietary work is kept on a private repository where only authorized programmers in Hong Kong and iCog Labs can access.
 
At iCog Labs various systems for their robots are in constant development including but not limited to CMT Tracker, Eye tracker, voice identification of a speaking person, gender classification from a speaker's speech and different object and situation recognition.

 
Hershey's

Hershey's or Hershey Foods Corporation is a chocolate manufacturing company founded by Milton S. Hershey in 1894. The company has been very popular for a long time now for the chocolates they manufacture. They have acquired many huge food and chocolate companies like Nestle and license to manufacture Kit Kat chocolates.
 
iCog has developed a Prediction Program for this company. The prediction program which uses AI techniques to guess the next thing from the example of the past is very important for a business company. With the ability to predict the future, although not realistically accurate all the time, one can make a lucrative business decisions based on the right tools and the right data about current and past events surrounding the predicted event. Even if the guess has more than half a probability of failing, it would be too important to ignore. That is why many companies have been looking for this type of technology. It could help an organization take fewer amounts of wrong steps.

 
Stevia First

Stevia First Corporation, Head quartered in Canada, is an agricultural biotechnology company that was launched in 2011 aiming to eradicate obesity and diabetes epidemics. They planned on undertaking this by trying to end sugar addiction. The company partnered with scientists and agronomists to supply the world with Stevia: a natural sweetener they call “perfect” and “The holy grail of sweeteners”.
Through the various research and development work they have done since their beginning, they came to one that involved iCog; the relationship between genes and longevity. They were interested in finding out the specific genes that contributed to diabetes. Particularly, they were pursuing “geroprotectors” which have been identified to have effect on age also with the presumed ability to prevent or even reverse metabolic disorder, obesity and Type 2 diabetes.
 
iCog Labs, as a company that has been dedicated to utilizing AI armed with a well competent bio-informatics team undertook the project. The team made use of an evolutionary learning algorithm to do analysis on the data using MOSES; a machine learning tool that genetically constructs and learns programs to arrive at the best possible one through evolution. Moses was developed by our partner, the OpenCog Foundation with Moshe Looks.

 
Build Direct

Founded in 1999, in Vancouver, the British Colombian Tech Company is an online marketplace for heavyweight home improvement products. As it is a common challenge for every big company with a vast number of customers, Build Direct faced the problem with customer support. It has thousands of customer service employees working 24/7 to interact with its customers and provide them with crucial information to address inquiries regarding products and services or help resolving customer complaints. Over 3.8 million kilograms of BuildDirect products are shipped oversea each month and its relatively smaller number of customer service officers gathers their information through a telephone call or chat dialogues with the customer.
 
Starting from January 2017, iCog is developing autonomous customer support software with an intelligent chatbot which is responsible to assist the customer service of BuildDirect. The chatbot we are building for this company is AI enhanced software that can converse with humans to answer questions. The BuildDirect chatbot will have the capacity to interact with the customer through a phone call or text based chatting and helps customers find what they wanted to buy as the human customer support officers do. Our system is capable of acquiring what knowledge a customer service could have by combining the pre-set texts (log from previous customer service and customer conversation) and AI learning algorithms, and generate response to a posed question or statement.

Opencog

iCog Labs works with the Opencog Foundation to create an open source AGI (Artificial General Intelligence) platform that could one day have intelligence equal to or more than the human brain. AGI is the better form of AI which is to create machine intelligence that is broad and not limited to a specific need. The AI system being used to date has mostly been task-oriented and implemented with a single application in mind. Game AI, object recognition and NLP are examples of task oriented machine intelligence. Hence, it is called Narrow AI.
 
With Opencog, the plan is to create an AI that learns, deduces and expresses itself in a simillar manner to humans, without the need for a specific goal being implemented individually. Dr Ben Goertzel and his team have already implemented the best parts of OpenCog; The Atomspace and Moses. The Atomspace is a knowledge bank for representing the knowledge the AGI agent has gathered in a way that allows the combination of different pieces of knowledge along with confidence and memory and attention span. Moses is an evolutionary program learner which based on some input data, designs programs that could understand the data and evolutionary (using concepts of evolutionary science such as natural selection and mutation) arrives at the best candidate programs.
 
The Opencog project is utilized almost in every project the iCog team undertakes. Aside from utilizing Opencog, the team also develops Opencog. Bugs are fixed, modifications are made and new features are added. Our work on the Opencog project is an ongoing development with lots of uncharted territory to discover.

 
SingularityNET

SingularityNET is a project aiming to build a network of AI services where the synergy of agents in the network would be able to evolve into a general intelligence. This network of intelligent agents is the accumulation of various AI services offered by individuals and organizations. The network would serve as a marketplace to trade AI services creating a fully decentralized and democratic access to AI and its offerings. SingularityNET is a blockchain- based framework designed to serve as the payment infrastructure of AI Agents interacting with each other and with external customers.
 
This project initially aims at inspiring AI developers around the world to upload their algorithms on the network and get paid from people using it. There is an AGI token which AI agents and humans will utilize for payment. Once there are enough agents in the system, including agents that make use of other agents to provide services, the network shall evolve into a giant, self- organizing agent on its own.
 
iCog Labs is in charge of developing this system along with a team of engineers in Russia, Hong Kong and various other countries in the world.

 
MOZI

MOZI is a project for a client in Hong Kong. The main goal of the project is to develop an AI powered genetic analysis tool for biologists. The project allows biologists to upload gene data, then it will analyze the data to provide any interesting analysis of the data.

The project is aimed at making big data analysis much easier for biologists working on Genetic Science. Gene data is too bulky to do analysis using traditional methods and it becomes necessary to use machine learning tools. However, it is quite obvious that a Biologist wouldn’t be as effective with an AI tool compared to an AI scientist.
 
The MOZI project utilizes MOSES for doing data analysis. MOSES, which stands for (Meta-Optimizing Semantic Evolutionary Search) is a supervised learning algorithm which uses evolutionary and genetic algorithms to learn programs. These programs are modeled after an input data which MOSES examines and evolves the best program to analyze the data. The best program is selected out of many candidate programs through a process that is much like natural evolution. A set of programs may be mutated and selected according to their fitness.

 
Community Projects

It's more than important for iCog to have close contact with the developer community in Ethiopia and beyond. With the aim of creating a tech savvy society, iCog has two community projects investing in the future. The company works with students and teachers aiming to create a better tomorrow where the youth of today become leaders and innovators.
 
iCog Makers Initiative
iCog Anyone Can Code (ACC)
iCog SolveIT

 
iCog Makers Initiative

iCog Makers has a signed agreement with more than 21 Ethiopian universities, and 2 public universities from Nigeria and Kenya to run this project in three phases. The very first phase is a training seminar/workshop on the concept of Artificial Intelligence and Robotics. We do not stop there! The iCog Makers Initiative has launched an annual tournament between public universities of Ethiopia, with the aim to reach out to the rest of the continent. The first phase of this tournament is called iCog Makers Robosoccer Cup. The initiative envisions technology as a conduit to deliver and enhance the formal curriculum taught in high-schools and universities through hacking competitions. These competitions will undoubtedly encourage creativity and technical development.

Through the employment of toy Robots, self-driving cars, and drones the iCog Makers Initiative provides the perfect methodology to teach students the concepts of humanoid robots, autonomous system, artificial intelligence, computer engineering, robot mechanics and many more subjects related to the science of hi-tech. Our initiative integrates some of the very important engineering concepts pertaining to mechanics and electronics. The annual competition involves fully autonomous robots, with no remote control, competing against each other in a 3-on-3 soccer. Each team uses the same kind of humanoid robots but via different micro-controllers and hacking designs. And, to the delight of our universities, the competition is focused on AI software. The universities form a team after the workshops we provide and train them to compete in the annual robosoccer cup tournament.
 
On the Second phase the universities buy the hardware and equip their teams with the right gadgets. After a 3 months development period, the third phase is launched; the actual robot soccer matches between the universities.
 
We plan to do more competitions and training sessions that have the power to enhance the minds of our youth so that they’re competent enough for the global market. We firmly believe that we should no longer lag behind in the technological arena!

 
iCog Any One Can Code

This is the same version of our Makers Initiative with a shifted focus towards a high school and elementary school children. We have launched an academy where children from age 6-10 are instructed with the basics of programming using simple languages and platforms like Scratch.
So far the iCog Any One Code (ACC) initiative has successfully delivered free trainings, workshops and weekly programming tips to more than 1500 highs school students in Addis Ababa and Bahir Dar city in collaboration with US embassy, Camara Education, Bahir Dar University and other stake holders.
In the academy under the ACC, we have been educating a dozen children under the age of ten in just three months with an expansion plan to reach 120 children per year.