R42 Institute

R42 Institute AI Fellows Program

R42 Institute  is running an AI Fellowship Program. A launchpad for deep technology and science disruptors, the Fellowship is designed to develop AI/machine learning, deep science, design thinking and entrepreneurial skills of emerging talent.


The Fellowship has been designed to provide each Fellow with one-on-one mentoring, an opportunity to hone their tech and science skills and combine it with design thinking tools over an accelerated period. 


This is an online program, with virtual interaction.


There is no fee for the program. The Fellowship is highly competitive and spaces are limited.


AI Fellows will receive:


  • R42 Institute AI Course (both live and recorded classes) - Fellows can get accelerated content to work on their projects

  • One-on-one mentoring with R42 team

  • All R42 lectures

  • R42 Institute Certificate

  • A unique opportunity to build a wide network of inventors, entrepreneurs and investors


Dates: Rolling 8-Week Program.


Application Process:


  1. Apply: Our application is simple. Tell us which project you are interested in, or propose a new one for R42, and a bit more background on yourself.

  2. Interview: After receiving your application, we will schedule a 20-minute online video interview with a member of the R42 team to learn more about your background and interests.

  3. Get Notified: After the interview, you will receive an admissions decision quickly - within 1-2 days. 


About R42 Institute:


R42 Institute, part of the R42 group, educates about AI, longevity, biotechnology. Its mission is to further define the university of the future which will be ever more important in the post-COVID world. This is another opportunity to seek ways to create the future and not just react to it.


Current Project Offerings: 


Each Fellow will undertake a project. Below are the projects R42 is offering. More projects may be listed later. Please, contact us directly if you have any ideas for projects for R42 and its portfolio companies.


1. Graphical AI Platform (with Dr. Ronjon Nag)


  • Artificial Intelligence (AI) principles are now widely used in industry and many disciplines, not just in computer science, but also in physics, engineering and chemistry.

  • Textbooks and courseware are now available explaining AI in easy to understand language. Over time the tools have become easier to use with libraries for Python.

  • Platforms such as Tensorflow and Keras have made it much easier for those who have a python programming background to implement quite advanced AI systems such as convolutional neural networks in just a few lines of code.

  • Notwithstanding this, graphical user interface platforms are few and far between, and we are far away from an “Excel for AI”. This project will start to put the framework for such a project to allow users to upload data, clean data and change neural network parameters from a graphical user interface, maybe creating an upgraded form of playground.tensorflow.org.

  • Pre-requisites: a Python programming background.


2. Stock Market Prediction (with Dr. Ronjon Nag)


  • Predicting the stock market has been the holy grail for financiers. There have been many mathematical models.

  • Most often, however, these models rely on their own past data. We will be taking signals from a model that does not use past data and attempt to get a model that takes an ensemble of predictive signals to find the best model for the trading system.

  • We will be working with signals from Algodynamix.com. Pre-requisites: knowledge or willingness to learn, a statistical programming language like R, Python, SAS.


3. Making AI Easier to Explain: Creation of AI examples for Use in Teaching (with Dr. Ronjon Nag)


  • This project will involve creating sample code in python on the google colab platform. The projects will take a problem, and have code that can upload a dataset, and allow users to change parameters in various kinds of neural networks, using the Keras platform.

  • Pre-requisites: Python.

4. MemLove - Mental Health App for Grievers (with Praveena Dhanalakota)

  • MemLove is a uniquely personalized proactive mental health app for grievers, especially now due to COVID-19 impact, to cope faster and healthier by emulating the voice of loved ones. MemLove's mission is to recreate and keep memories of loved ones alive through artificial intelligence.

  • We hypothesize that the application of psychology models such as Continuing Bonds can impact ~100M people yearly in the US and defined social benefit capabilities. We are in the phase of system and services design to devise a framework that provides suitable and sustainable ways to build and deliver the services to users.

  • Some of the technology areas we are focusing on are:

    • Speaker diarization

    • Voice emulation

    • Speech emotion recognition

    • Knowledge graph for predicting possible conversations

  • Student Goals:

    • Experiment on speech sentiment analysis

    • To analyze voice data to understand emotion and sentiment.

    • Experiment on the knowledge graph

    • Recreate mannerisms based on past conversation data available. This will enable to frame interactive and therapeutic responses

    • Author a blog post for experiments

  • Pre-requisites: Programming experience (minimum 1-2 years)


5. Chatbot Creation for 5G Messaging and AI (with Dr. Ronjon Nag)

  • Chatbots are prevalent, but how do we make them easily. The idea of this project is that it is a subset of the AI Graphical UI project. The idea is to be able to create a chat bot by uploading chats and creating a chat model.

  • Microsoft Teams could be used as the basis for the development of the system.

  • R42 has a portfolio company: which has the Ecrio nimbus 5G Messaging Gateway to reach out to Smartphone users using Ecrio 5G Messaging App.

  • Possibly could be used in a hospital context where patients with their consent can use 5G Messaging App to communicate with ER staff in the hospital via ER-Bot. 

  • The 5G Messaging App can be extended to leverage the Covid-19 Contract Tracing APIs (available on Android and iOS for developers) and feed that information to the ER-Bot.

  • ER-Bot uses either Azure Cognitive Services or any 3rd Party AI/ML frameworks to model the conversations among the ER-staff as well as with Patients. The model needs to be compiled, and trained before it can be put to use for predictions either in the simulated phase or the live phase of the project.

  • The team however, would be expected to research the problem first before any building, with interviews, and use a design thinking framework to prioritize high impact projects. Some possible ideas could be:

  • Potentially a proof-of-concept could be built using simulated ER and Patients and Simulated Data.

  • Maybe a contact tracing chat bot could be implemented. There are several open source contact tracing frameworks.

  • Ideally this should be built as a module in AI Graphical User Interface.

6. Water Data Science (with Russell Perry)

  • California has a complex water system, from widely varying ecosystems for water production to water consumer end-uses. It recently experienced a large drought (2014-2016) and may be entering another. Drought events have had a huge impact on water behavior, breaking many older water models by changing the assumptions behind them. Additionally, COVID-19 has drastically impacted water behavior, moving much of the water end use to residential locations rather than in businesses, further complicating water district’s models.

  • Water retailers have millions of rows of historical data on production and consumption, and new infrastructure in the form of Advance Metering is being installed in several districts giving access to very granular water data at the end use.

    • How can this data be best used? 

    • Can new behavioral models be created?

    • Can the data be used to assess leakages? 

    • How can older mechanical water use and water savings models be remade with a more dynamic basis?

  • At the water production level, retailers, wholesalers, and other water organizations rely heavily on using weather forecasting with watershed models to predict incoming water volumes. \

    • Can this process be improved? 

    • Can it be integrated with an end use model for an “all-in-one” model?

  • Water data science is not as far along as many other industries, and this is in part due to a lack of data centralization, data is hard to get.

    • How can we improve this, and bring the data together?

  • The water industry is ripe with data questions and is a prime opportunity to use what has been learned in other data arenas and apply it to a new set of problems. As final question, expanding out of California, can these processes generalize?

7. Protein Folding (with Berke Buyukkucak)

  • Summary: Predicting protein folding is the holy grail of biology and there has been great strides made in this field, thanks to machine learning. Fellows that participate in this project will decipher, understand and try to improve the AlphaFold code that was implemented and then made public by DeepMind, Google's AI company. 

  • Motivation: Protein folding prediction knowledge and skill can be translated into machine learning driven drug discovery research, which is a growing and forceful field that shows the potential to revolutionize both pharmaceutical and medical research. 

  • Challenges: Difficult concept to understand and implement, terminology based knowledge.

  • Impact: Learn about a unique field that's developing, get students educated and knowledgeable.

8. Graph Networks for Physics Simulations (with Steve Crossan) 

Recent work in Graph Neural Networks has demonstrated how this class of machine learning architectures can be used to learn efficient representations of physical simulators in a variety of domains. This project will use DeepMind’s open source Graph Nets framework to 1) replicate the Water simulation (SPH) and + 2) attempt to produce the same result using a differentiable simulation framework, either Jax-MD or Molly.jl 

9. Deep Science Business (with Anastasiya Giarletta)


  • This role will assist in R42 development activities, evaluating companies, reaching out to R42 portfolio companies and facilitating assistance to companies either in direct research or cross-company synergising.

  • Will write a business report after each R42 lecture for publication on Medium, company website.

10. Ecosystem Development and Growth Marketing (with Anastasiya Giarletta)


  • This role will assist in R42 Institute development activities.

  • Utilize digital marketing tools, including analytics.

  • Optimize marketing campaigns for R42 lectures and courses.

  • Design R42 decks and website.

  • Manage all social media platforms and drive content creation.

  • Write a blog post after each R42 Institute lecture for publication on Medium, company website

11. Mathematics of Venture Investing (with Dr. Ronjon Nag)


  • Venture capital investing is a risky business and there are many proposals on how to do it to optimize. returns and reduce risk. Some views say invest in a few companies and work deeply with them. 

  • Others say that the distributions are highly skewed and therefore one should invest in lots and lots of investments to capture the high growth outliers.

  • However, this means that the best entrepreneurs may not want your investment since they demand your time and full attention. Some say invest only at the earliest stage, since the returns will be higher, but of course this is the riskiest stage.

  • Others say invest in late stage companies when the risk is taken out.

  • The objective of this project is to review the current models, and suggest new models for R42.

©2019-2020 by R42group.com, a Thinkica Company