At Insight, we want to make sure that everyone with the right skills can participate in our programs, regardless of financial challenges. So, in addition to offering several scholarships to help reduce the barriers that keep you from taking those first steps to launch a thriving career, we are now offering deferred payment loans to Fellows in all of our programs through a partnership with Climb Credit. These loans are offered through Climb Credit’s upfront financing, and are financially backed by Insight to support our Fellows during their career transition.

Climb Credit Financing

Climb allows Fellows to make manageable monthly payments over a few years. Their loan process is simple and fully-online, and the vast majority of applicants receive a funding decision almost immediately. This application process doesn’t do a hard credit check, so applying won’t have a negative impact on your credit score. Climb works with people from all kinds of credit backgrounds, including those without any credit to their name.

Climb Offers:

• A quick online application, which can be completed in as little as 5 minutes with no impact to credit
• High loan approvals — including financing for people with no credit
• Instant decisions 90% of the time, with the ability to accept and e-sign your documents in just a few clicks
• Climb allows you to set up auto-payments, so you don’t have to remember to make your payment each month. As an added bonus, they will reduce your interest rate by 0.25% if you sign up for autopay.
• You can pay off your loan early without any pre-pay penalties
• No payments until 9 months after you join Insight, ensuring you meet all components of the Insight guarantee

Backed by Insight’s Guarantee

As with all of Insight’s financing options, the Deferred Payment Loan option is backed by our guarantee: you will only complete the financial commitment if you achieve the high level of success that we expect for our community. When you select financing through a Climb loan, you only pay if you accept an offer for a relevant job earning at least $100,000 (and that offer is dated within 6 months of the end of the program).

Helping our Fellows into thriving careers

Ultimately, all of our partnerships and initiatives have one goal — the long-term success of our Fellows. We’ve helped more than 3,000 Fellows start thriving careers at over 700 top teams. 88% of our Fellows join relevant roles within six months of joining the program, and we continue helping every Fellow until they find a good fit.

And their success continues well beyond the 7 weeks of our program. The same strong drive and collaboration that makes our Fellows successful in our program helps them become leaders in their fields. Our Fellows lead teams at industry pioneers like Amazon, Apple, Bloomberg, Capital One, CVS, Facebook, Google, LinkedIn, Lyft, Microsoft, Netflix, Pinterest, Spotify, Stitch Fix, Tamr, Twitch, Twitter, The New York Times, Uber, Wayfair, and Yelp. Our Fellows also lead teams at hundreds of high-growth startups, and several alumni have gone on to start their own innovative startups.

The most rewarding part of our program, and the reason Insight exists, is the experience we provide for our Fellows. We’re incredibly excited to continue helping our Fellows transition to thriving careers with a growing number of financing options.

Interested in transitioning to a thriving career? Learn more about the Insight Fellows Program and start your application today.

* While Insight offers all of our other financial aid options to all Fellows, Climb Credit can only offer financing to US citizens or Permanent Residents.

Insight now offers deferred payment loans was originally published in Insight on Medium, where people are continuing the conversation by highlighting and responding to this story.

The Remote Experience

While this was the first session for Insight in which all Fellows participated remotely, we’ve been offering a remote data science program since 2015. We’re happy to report that we successfully adapted what we’ve learned during that time to all 7 of our programs across all locations. While certain aspects of the remote experience looked a bit different than the on-site offering, many remained the same, and there are also several benefits to a remote program that a more restrictive in-person session simply can’t offer. Insight continues to provide Fellows with full support from our teams:

• Program Directors work directly with Fellows to provide expert mentorship, advising on project selection, development, and demoing with hiring companies.
• Program Operations Team provides expertise for managing the crucial program operations (communications, planning, logistics, tracking, etc.) necessary to ensure the success of each Fellow.
• Coaching & Development Team provides high-quality training, coaching, and feedback via workshops and small group sessions, helping Fellows identify the best roles at this stage of their careers and build the self-awareness and confidence to communicate their unique value-add.
• Interview Strategy Team is fully dedicated to the post-session experience, providing support and mentorship services through dedicated training and coaching groups in service of helping Fellows build the bridge to thriving careers.
• Partnerships Team works directly with hiring companies to understand their specific needs and secure exciting opportunities for our Fellows across all programs and locations.

Fellows benefit from the same small cohort sizes that the in-person Fellowship offered. None of our program cohorts have grown beyond 30, with the average being about 20 Fellows. Technical advisor and alumni mentoring meetings provide opportunities for even smaller groups to gain feedback and advice, and personalized interview preparation is conducted on a one-on-one basis, tailored to relevant skills required for that company, and based on the existing skills of the individual Fellow.

Fellows now also enjoy the technical advantages of using industry-standard distributed tools such as presentations via Zoom, internal messaging platforms (like Slack), and Github to contribute to a shared codebase. The remote session enabled Fellows’ increased flexibility and a more expansive reach. This summer, Fellows reported that they enjoyed the ease of meeting with other Fellows and Insight staff members every day without the wasted time of having to commute. Participating remotely also meant that Fellows were better able to set their own work times, instead of the more strict 9:00am — 6:00pm that is required during the in-person session. The remote experience made it much easier to connect and work with Fellows from other locations, and expanded access to Industry Leader Mentor presentations that are generally restricted to Fellows located in the city where the mentor visit is hosted.

Industry Leader Mentor Sessions

An exciting aspect of Insight’s Fellows Programs is the opportunity to participate in Industry Leader Mentoring Sessions, in which some of the tech industry’s top professionals visit with Fellows to speak about their experiences, share perspectives, and answer questions.

hey @jakeklamka, thank you for having me last week for an AMA with your @InsightFellows cohort! pic.twitter.com/6QEL8WW1yH

During the summer session, Insight was proud to host an impressive lineup of industry leaders, which included:

Alexis Ohanian, the co-founder and managing partner of Initialized Capital. He was the co-founder, and later Executive Chairman, of Reddit. Ohanian is also the best selling author of Without Their Permission.

Peter Norvig, a Director of Research at Google. He was previously head of Google’s core search algorithms group, and of NASA Ames’s Computational Sciences Division, making him NASA’s senior computer scientist. Norvig is co-author of Artificial Intelligence: A Modern Approach.

DJ Patil, who served as the first Chief Data Scientist of the United States Office of Science and Technology Policy, and is currently the Head of Technology at Devoted Health. Patil is credited with coining the term “data science”.

Hilary Mason, the co-founder of Hidden Door, founder of Fast Forward Labs, a machine intelligence research company, and the Data Scientist in Residence at Accel. Mason was previously the Chief Scientist at bitly and co-founder of hackNY.

Solomon Hykes, the Founder, Chief Technology Officer and Chief Architect of Docker and the creator of the Docker open source initiative. In his role at Docker, Solomon is focused on building a platform for developers and system administrators to build, ship, run and orchestrate distributed applications.

Drew Conway, the Senior Vice President at Two Sigma. He was previously the founder and CEO of Alluvium, and is a leading expert in the application of computational methods to social and behavioral problems at large-scale.

Chris Wiggins, the Chief Data Scientist at The New York Times, and an associate professor of applied mathematics at Columbia University. At Columbia, Wiggins is a founding member of the executive committee of the Data Science Institute, and of the Department of Systems Biology. He is also a co-founder and co-organizer of hackNY.

Anne Bauer, the Director of Data Science at The New York Times, leading the team in charge of algorithmic content recommendations. Bauer is also an alum of the 2015 Insight Data Science Fellowship Program.

Wes McKinney, an open source software developer focusing on data analysis tools. He created the Python pandas project and is a co-creator of Apache Arrow. McKinney is currently a director at Ursa Labs, a member of The Apache Software Foundation, and a PMC member for Apache Parquet.

Diversity, Equity, and Inclusion Summit

This session, Insight hosted a Diversity, Equity, and Inclusion (DEI) Summit, with the goal of facilitating discussions to share insights, challenge perspectives, pose stimulating questions and build community with each other so that Fellows are better able to enact positive change in their new roles. This event brought together Insight Fellows, staff, and thought leaders from our community to discuss a range of social issues that impact how we work together, and society at-large.

Lightning talks from Insight staff, an alumni panel, and a keynote speaker were utilized to spark conversation for discussing important — and sometimes difficult — topics that we don’t often have the chance to talk about candidly in our day-to-day work; topics that nevertheless tie into how we experience the workplace and interact with each other.

The event Included:

Lightning Talks with Insight Program Directors, who discussed a toolkit for recognizing and addressing your unconscious bias, and reflecting on anti-Black violence.

Alumni Panel with Insight alumni Melecia Wright & Che Smith, who discussed the role of companies and technology in building a more equitable and socially just future.

Keynote Speaker Brandeis Marshall presented, Deepfake Technology: The (Mis-) Representation of Data. In this session, Dr. Marshall described deepfake technology and discussed issues related to ethics, privacy, and accountability for companies who use “black box” algorithms to make decisions and the respective societal impacts.

As Fellows prepare to enter the job market and join a professional community, we want to help them think critically about how they can contribute to creating more diverse, equitable, and inclusive work environments. Below are a few reactions that Fellows shared about the event:

• “Thank you to every person for exposing your vulnerability to such a large group. It really helped me to engage with the conversation because I felt safe there.”
• “…I have attended many diversity training workshops in my schools and my previous job. However, this was the best and most genuine of them all. I can see all the effort put in it. Thank you!”
• “I am so grateful to Insight for letting us pause for a moment and think about our mission and responsibilities to make this world a better place through science, tech, education, communication, and reaching out to the congress to hear our voices. And most of all, I found a group of inspiring women during our discussion time through Insight platform.”

Fellow Spotlights

We have been working with so many incredibly talented Fellows this summer, and spotlighted a few to reflect their wide range of backgrounds, special skills, and experience.

Jayakrishnan Parappalliyali, DevOps Fellow from San Francisco. At Insight, he built an autonomous stateful application in a cost-efficient cloud infrastructure.

Paige Chang, Artificial Intelligence Fellow from NYC. During the session, she applied a cutting-edge approach (reinforcement learning) to a classic data problem (song recommendation), all deployed using AWS and Tensorflow.

Cameron Jones, Data Science Fellow in San Francisco. Cameron built an app for streaming positive, relevant news for Black audiences by scraping and processing articles through Natural Language Processing (Google’s BERT model).

Saba Khalid, Data Engineering Fellow from New York City. For her project, Saba built an efficient data pipeline to ETL Federal Elections Committee (FEC) data to map changing trends in individual campaign contributions.

Karen Larson, Health Data Science Fellow from Boston. Karen has been consulting for University Hospitals, building a tool to help doctors avoid ordering unnecessary lymph node biopsies for melanoma patients.

Javed Jaghai, Data Science Fellow based in Washington D.C. Javed leveraged a deep learning RNN model to build the world’s first two-way translator between English and Jamaican Creole.

Justin Kaseman, Decentralized Consensus Fellow based in the Bay Area. At Insight, Justin built a tool to easily bootstrap and deploy a decentralized application to help bring new developers into the space.

Doaa Altarawy, Data Science Fellow in Toronto. As a Fellow, Doaa applied her strong data and engineering skills to an incredibly timely project: screening chest x-rays for quick detection of COVID-19, using fastai’s deep neural net model.

Sarah Uludag, Data Engineering Fellow from Los Angeles. Sarah built a tool for finding and marking public keys used in Bitcoin transactions on the Dark Web using Apache Spark, Cassandra, and Neo4J.

Post Session Experience

The initial 7-week program of the Fellowship has come to an end, but Insight’s work with Fellows has only just begun. Fellows spent the session building projects and demoing their work, and are now engaged in interviewing with hiring companies. As of August 28, there are 356 opportunities from 238 companies, and that list continues to grow. This is a difficult time for the economy, but we’re happy to report that our current growth rate of hiring opportunities is on par with our summer session in 2019. Companies currently interviewing Fellows include Netflix, Amazon, Facebook, Google, Yelp, Apple, Pinterest, The New York Times, Johnson & Johnson, Humana, CVS, ProtonMail, Bolt Labs, and many more!

Insight’s Interview Strategy Team is currently working with our Fellows on an individual level to provide personalized interview preparation support and mentorship services. We’re investing in our Fellows to honor our guarantee and ensure they’re hired quickly to launch exciting new careers in tech.

The Insight Guarantee

Looking ahead to our Fall 2020 Session

Insight is currently selecting our next Fellow cohort for the fall session, and continuing to work to help reduce the barriers that keep potential Fellows from participating in Insight’s programs, taking those first steps to launch their thriving careers.

Scholarships

We’re expanding our current offering of scholarships beyond our Need-based Scholarship, which has been available for several years to help fund essential day-to-day living needs related to the program. Starting this fall, we will now offer two additional scholarships:

• Scholarship for Underrepresented Minority Groups — This fund has been established to help remove barriers for our Fellows from racial and ethnic backgrounds that are traditionally underrepresented in tech. Eligible Fellows will have the opportunity to complete a simple application process in order to be considered, and those selected will receive up to $3000.
• Gender Diversity in Tech Scholarship — Insight is proud to continue our partnership with Clover Health, who is sponsoring this scholarship fund to help remove barriers based on gender in the tech industry. Eligible Fellows who complete the application process will be considered for a $5000 scholarship.

Mentorship Opportunities

This fall, Insight is also piloting a mentorship program for candidates who identify as one or more historically underrepresented groups in tech (American Indian, Black/African American, Hispanic, Latinx, or Native Hawaiian/Pacific Islander). The purpose of this mentoring program is to provide an opportunity for applicants to ask questions about the interview process, program, post-program experience, and workforce more broadly. Many applicants do not have existing connections to the Insight network that they can turn to for information that may be available to other applicants. As we learn from this pilot program, we’ll continue to improve upon the offering with the goal of providing a more inclusive experience for all.

Insight’s core value is to put Fellows first, and we are, first and foremost, committed to the long-term success of our Fellows. As we plan for the future, we’re excited to not only reach the same impressive outcomes, but to exceed them. We’ll continue to set a high bar for what constitutes success within the program, and maintain the guarantee of a job in a relevant field, earning at least a $100,000 salary within 6 months of the end of the program. In addition to our guarantee, we’re committed to improving upon and expanding our accessibility to an even broader audience.

Are you ready to make a change & transition to a career in tech? Sign up to learn more about Insight Fellows programs and start your application today.

This post was revised on August 28, 2020 to update the number of company opportunities.

Insight Summer Session 2020 Update was originally published in Insight on Medium, where people are continuing the conversation by highlighting and responding to this story.

While this conversation has surfaced from the most recent events, we acknowledge that it is set against a backdrop of systemic discrimination and an ongoing series of injustices. We as a community must continue to stand up, call out these injustices, and work together for the foreseeable future to address this problem in our society. We stand in solidarity with the Black community, and those who are protesting against systematic oppression.

We’re working to create an open dialogue within our team and Fellows on this difficult topic. We’d like to share a few resources that are helping our team, our Fellows, and our larger Insight community.

8 Can’t Wait

Data-driven Campaign Zero suggests eight policies that can decrease police violence by 72%. Learn more and take action in your own city.

Rage, Fear, and Confusion

DJ Patil, former U.S. Chief Data Scientist, has long been an advisor to Insight and continues to be someone we look to for thought leadership. Over the weekend DJ shared some of his thoughts and reactions in this Medium post. In it, he shared his view of the specific policies we should be accelerating on now, where data and technology can help drive the change we need now.

Why CEO Black Lives Matter Communications Are Critical: A DIBs Leader’s Perspective

Erin L. Thomas, VP and Head of Diversity, Inclusion & Belonging at Upwork, shares anti-racist corporate commitments companies should be considering.

How to Manage Your Team in Times of Political Trauma

Michelle Kim, Co-Founder & CEO at Awaken and Diversity, Equity & Inclusion Advocate, shares some thoughtful advice on what companies should do to empower their managers to “actually behave as inclusive and equity-conscious leaders.”

Anti-Racism Resources

This anti-racism resources Google doc provides a list of articles, videos, books, and other resources to help us better engage in anti-racism work. Author Ibram X. Kendi also collected this anti-racist reading list.

100 ways you can take action against racism right now

For those who are looking for ways to get involved, this article provides a list of recommendations from contacting state and local leaders, making donations, volunteering your time, supporting businesses and much more.

Acknowledging and discussing the recent, tragic events was originally published in Insight on Medium, where people are continuing the conversation by highlighting and responding to this story.

Insight is launching the Insight Security Fellows Program, a professional training fellowship to help systems engineers, software engineers, and data analysts transition into full-time careers as security engineers, data privacy engineers, and advanced security analysts. This free, full-time, 7 week fellowship based in San Francisco will start in September 2019. Fellows in this program will build secure applications and platforms using the latest technologies in the security ecosystem, with mentorship from leaders in the security field. Following the 7-week program, Fellows will interview for full-time roles at leading security teams in Silicon Valley, receiving guidance and support until they accept a full-time offer to join a top company.

Trust as the foundation of the modern tech stack

Our apps and connected devices produce a deluge of data that is collected with the goal of improving and funding the services that enrich our lives. While this model fuels many of today’s businesses on the internet, it comes with a significant tradeoff: an unprecedented amount of user data has been stock piled and is at risk of being exposed through security breaches. Thus, the development of products and services that leverage user data brings a new ethical responsibility for today’s tech community: securing the infrastructure for the services we use everyday, and protecting the privacy of our data.

Unfortunately, the concerns over lax security practices have been vindicated far too often in recent years. Every record of personally identifiable information leaked erodes the trust of users, risking the value of new features that teams work so hard to build. With the latest round of security breaches and data leaks, we’ve reached a tipping point where users will no longer tolerate the tradeoff between privacy and free features. If a company treats security and data privacy as afterthoughts, then users will vote with their clicks and take their data elsewhere.

As a result, companies need to ensure security and privacy are core components of their products. Engineering teams need to prioritize preventing security breaches as much as they focus on new features or code quality. If your users don’t believe that you’re looking out for their best interest, every feature and application you build could be jeopardized— trust must be at the foundation of the modern tech stack.

Fortunately, many companies are taking on these challenges with new teams dedicated to protecting their users in their security orgs and beyond. Security engineering teams are leveraging a new technologies in cloud automation and data engineering to defend against security vulnerabilities at a new scale. InfoSec teams dedicated to anti-abuse and data privacy are using recent advances in machine learning and engineering to help secure their services. Data stewardship programs are thoughtfully considering their use of data, ensuring that collected information is truly useful to the end user. A new field of data ethics is forming amongst industry leaders, with the goal of actively ensuring algorithms don’t discriminate against protected classes or have deleterious effects on society.

While there’s a huge demand for security engineers and analysts with this new skill set, there are few with the required knowledge in the security and data privacy domain. More and more companies are realizing that their future depends on their ability to earn the trust of their users, and a quality security team will be a key competitive advantage to their long-term growth. They’re searching for a new generation of thoughtful and talented security professionals, but are struggling to find enough of those individuals with the relevant skills.

For years, organizations have tried to invest more in security, only to face a lack of qualified engineers and analysts. Industry experts have been concerned about the shortage of security professionals, expecting the number of unfilled positions to increase to 3.5 million by 2021, a more than 300% increase since 2014. Despite urgent calls for more education and training from leaders — from computer science leaders to Fortune 500 executives to the White House — the demand has far outpaced the supply. The skills gap is even sharper in the advanced security roles that require knowledge of data, cloud infrastructure, and machine learning skills on top of the security domain expertise.

At the same time, there is a healthy contingent of engineers, IT professionals, scientists, and analysts who are passionate about data privacy and securing the internet from bad actors. This group has 90% of skills they need to succeed in the field of security— with the necessary fundamentals in software and systems engineering — but need the hands-on experience with the contemporary tech stack that security teams are looking for. Despite the potential of these candidates, many teams can’t afford to take a chance for such a critical role — they want to see relevant experience working in the security domain.

This is the mission of the Insight Security Fellows Program — to bridge the gap between ambitious professionals with a passion for security, and the companies looking to do the right thing for their users.

About the Security Fellows Program

Every Fellow who completes the program will build on their foundation, and come out of the 7-weeks knowing fundamental security topics like penetration testing, networking protocols, scripting, identity access management and malware defense. Fellows will join a network of ambitious software engineers, systems engineers, IT professionals, and data analysts who share common goals, but wield a diverse set of complementary skills. Fellows learn the same way as in the industry, by collaborating and working through real-world challenges with peers and experienced alumni who solve related problems in their industry roles.

Additionally, each Fellow gains an advanced specialization in security by completing a substantial project during the course of the program on advanced topics like cloud infrastructure, automation, microservices, container orchestration, user-level and ephemeral encryption keys, compliance with laws like HIPAA, CCPA, and GDPR, anomaly detection for social engineering attacks, homomorphic encryption, and differential privacy.

These projects will serve as the forcing function to learn the state-of-the-art practices in security engineering, advanced security analytics and data privacy. After completing the project, Fellows will interview and get offers for full-time roles at top security teams across the industry. Insight Security Fellows will have the opportunity to meet mentors from and be considered for roles at top security companies, including: LinkedIn, Stripe, Dropbox, Mozilla, Square, ProtonMail, Vectra, and many others.

The cohort will be made up of a diverse set of Fellows from a wide variety of backgrounds. The most effective security professionals have a personal drive to secure the internet and protect the privacy of users. They have a passion for understanding systems, thinking about failure modes, and building novel solutions to close vulnerabilities. While many Fellows will come from previous roles as systems engineers, IT professionals, software engineers, and data analysts, there are many paths to Insight. There is no educational requirement, and we’re excited to consider any creative applicant with a passion for security and data privacy, regardless of their background.

Like all Insight Fellowships, this 7-week program is provided completely tuition-free for participants through support from partnering companies. Participating Fellows will meet dozens of teams from industry, leading directly to interviews with top teams. Additionally, all Fellows will have access to full-time office space, cloud computing resources, as well as need-based scholarships and financial loans to cover the cost of living in the San Francisco Bay Area.

Since 2012, Insight has helped over 2,000 Fellows transition to cutting-edge careers in specialized technical fields at more than 700 companies across the US and Canada. Starting with data science, we’ve expanded our community to data engineers, machine learning researchers and engineers, product managers, devops engineers and blockchain engineers. We’re excited to add to this diverse network by helping the next generation of professionals join the critical and exciting field of security and data privacy. Learn more about and apply now for the Insight Security Fellows Program.

The first session of the Insight Security Fellows Programs in Silicon Valley will begin on September 9th, 2019.

To learn more visit the Insight Security Fellows Program, read the white paper, and apply today to be considered for the first cohort.

Insight launches Security Fellows Program was originally published in Insight on Medium, where people are continuing the conversation by highlighting and responding to this story.

Jared Yamaoka was an Insight Data Science Fellow in the Summer 2017. During his time at Insight, Jared built a machine learning model that used satellite images of Austin, TX to measure change in land use over time. Prior to Insight, Jared was a postdoctoral fellow at both Duke University and at the University of Hawaii at Manoa, and received his PhD from Rutgers in physics. Jared is now a Senior Data Scientist at GEICO.

This blog was originally posted on July 11, 2017 on Jared’s personal blog.

Recently there has been interest in using satellite images as investing tools. Hedge funds are looking at construction in Beijing to bet on concrete demand, or they are counting cars in Walmart parking lots to get an early estimates on profits. Here I discuss a project to determine land use (i.e. is an area a building or not a building) from satellite images. The idea was to measure the change in land use over time as an economic indicator. This project was a proof of concept for the Insight Data Fellows Program.

Images from the Landsat 8 satellite are readily available from several sources, including Amazon Web Services (AWS). The landsat-util program is a convenient way to search for and download these images. For my test case I chose Austin, TX. Austin has been the 2nd fastest growing city in the USA by percent of population over the last five years and has relatively clear skies. Clouds are the bane of satellite image analyses.

For this project, I also used the Quilt package manager. It enables the creation of data packages that are sharable and portable. More specifically Quilt provides data wrapped in a Python module as well as a repository for the data, a-la github. I downloaded and processed 4 images of Austin and built a Quilt package.

Quilt

Using the Quilt package is quite simple. The installation instructions can be found here. To follow along this tutorial you can check out my data package with all the images and labels you need to get started.

$ quilt install jared/landuse_austin_tx

From here you can browse the package in Python:

>>> from quilt.data.jared import landuse_austin_tx
>>> landuse_austin_tx
<PackageNode '/quilt_packages/jared/landuse_austin_tx'>
images
labels
metadata

I provide the original images as well as images cropped to a region around the city. These crops are much more manageable at 2 mega pixels (MP) than the computer crushing 60MP of the full images.

For my training image, I have chosen to use LC80270392014022LGN00 which is from 1/14/2014. A description of the naming convention can be found here. This was chosen because of the low cloud cover. To access the image containing light with wavelengths 0.630–0.680 µm (red band):

>>> b4 = landuse_austin_tx.images.cropped.LC80270392014022LGN00. \
... LC80270392014022LGN00_B4
>>> r = skimage.io.imread(b4(), as_grey=True)

More information about the available image bands is here.

Feature Engineering

The data was loaded and some higher level features were engineered to help better train the classifier. All bands have a resolution of 30m except band 8 which records all colors (panchromatic) at a resolution of 15 meters. This image was used to sharpen the red, green, and blue bands (r,g,b) by replacing the intensity in each band with the panchromatic image.

The above code also produces the normalized difference vegetation index(ndvi), which I call a “tree finder”. The ratio of reflected near-infrared light to red light is enhanced for plants (vs buildings and concrete).

NDVI = (NIR-Red)/(NIR+Red)

This is a well know feature in the remote sensing community and was suggested by a colleague (one of the nice parts of the Insight Program is the diverse background of the Fellows). The feature is very powerful and gives ~10% improvement in accuracy over a classifier trained with out it.

The second feature I call a “building finder” is designed to find edges in the image, and is known in image processing lingo as edge detection.

Each RGB band is blurred with a gaussian filter and then subtracted from the original image.

Data Labels

The labels for the training came from Open Street Maps which provides crowd sourced land use data. The vector shape files as well as rasters for the cropped images are available in landuse_austin_tx. The labels were a bit of trouble actually. The data available covers only 15% of the image and is highly biased to buildings. This leads to a strong over classification of “nature” as “buildings,” particularly in the farmland on the east of the city. To address this, the “nature” labels are interpolated as an area not labeled as a building. Training over a small fraction of the image and cross validating against the rest, there was less than 1% variation in accuracy. However labels are still one area that could be improved… most likely with some effort hand labeling.

Modeling

I modeled this as a two class problem, “Buildings” and “Nature” (anything not a building). In the future one can imagine separating out road or bodies of water (both considered “Nature” in the current model), but for this proof of concept this was not done.

Prior to large scale training, I moved to a c4.8xlarge EC2 container on AWS. This provided the resources, more specifically the memory to quickly explore the data and trained my models at least 20 times faster than my laptop. Here the Quilt system ensured the data was completely transferred and my code ran without having to change hard coded paths.

TPOT

I attempted to optimize the learning with the use of TPOT automated machine learning tool. This tool explores different models and returns the one with the best metric. It even returns the code to implement the model.

After running on a large EC2 instance for about 8 hours TPOT provided an optimized model, with code!

The default metric of optimization is “balance accuracy” which is an unweighted average of all class accuracies. This actually was not the best metric for this analysis, where the recall of buildings provides the best measure of the measure of the rate of urbanization.

I would love to re-optimize with TPOT using my preferred metric, but because of time constraints, I did an ad-hoc optimization and settled for an XGBoost classifier with 350 estimators. This returned a similar accuracy, but with a greater recall for the buildings (class 1).

Finally you can see the output of the classifier.

Time Series Analysis

Having trained a model, I began to look for changes over time. Here you can see the the output of the classifier for another image (LC80270392017030LGN00) from January, 2017.

Taking the classifier outputs from the 2014 image and 2017 image (shown above), I apply a gaussian filter to smooth the expectation and subtract. This gives the class change over time (red areas below).

As you can see I am able to find new construction (the red area in upper portion of the two images).

However the model is also sensitive to color or exposure variation that is not consistent with new construction (the red area in the lower portion of the two images). I believe more training across different images should help mitigate this effect, but more study is needed.

Conclusions

I have shown that I can successfully use ML to classify land use on a per image basis. The example mentioned in the Modeling section was trained on 10% of the data, but still achieved 83% accuracy (total) and 58% recall (buildings) across the entire image.

While the time series analysis will require cross image training or perhaps more feature tuning, the per image performance is reasonable. A more careful preparation of the labels and training several images is the next obvious step, but due to time constraints this wasn’t possible at this time. However those wanting to test the procedure for themselves, the Quilt data set and sample code are just a download away.

How to use Machine Learning and Quilt to Identify Buildings in Satellite Images was originally published in Insight on Medium, where people are continuing the conversation by highlighting and responding to this story.

Can technology give us the formula for healthier romantic relationships? Isaac Perron, a neuroscience PhD from the University of Pennsylvania tried to answer this question with his Insight project. He worked with Happy Couple to predict relationship health from over 8,000 users’ app data. Isaac is now a Data Scientist at WorldPay.

This blog was originally posted on February 13, 2018 by Happy Couple.

Is there really a way to figure out the secret to healthy relationships? The truth is, there are many answers — some will say communication, or supporting each other, or a good sex life. But it would take a detailed breakdown of the elements of each unique relationship, and a way to measure what “healthy relationship” means, to really offer a formula.

In the first study to show results on relationship health, research conducted by behavioral neuroscientist and current fellow at the Insight Data Science Program found that knowing your partner is more important for relationship health than being alike. Communication was also found to be one of the highest contributors to relationship health scores.

“…knowing your partner is more important for relationship health than being alike. Communication was also found to be one of the highest contributors to relationship health scores.”

The data science study by Isaac J. Perron, Ph.D. analyzed responses from an anonymized pool of 8,302 Happy Couple users in relationships who had answered all four couple health questions. The data allowed Perron to delve deep into correlations between users’ responses to all kinds of questions, and how they self-evaluated their relationship health in their answers.

“We’re thrilled to see that our mission to help the world have better relationships is working,” said Happy Couple founder and CEO, Julien Robert. “A good relationship is about asking the right questions to know and accept your partner, and we created this app to give couples a fun way to do this.”

One of the other most important factors to determine relationship health was how often users played the app. Perron found that the more questions a couple had answered on Happy Couple, the higher their relationship health scores were. This correlation could be for a number of reasons:

“While it’s certainly possible that playing Happy Couple makes relationships happier and healthier, the direction of causality could be reversed. Maybe happy couples just like to play Happy Couple?” Perron writes.

HOW THE STUDY WAS DONE:

(Click here for the full study)

Perron worked with relationship therapist, author and Happy Couple co-founder Dr. Lonnie Barbach to choose four ways to measure “relationship health”(or ground truths). They were how people rated their relationship (in the app quiz questions) in the four areas of: discouragement, sex, communication, and fighting.

From there, he analyzed how couple’s health scores in these four health categories were affected by being the same(alignment) and knowing each other (connection) in each of the app’s 6 quiz-questions categories. The app’s 3,000 questions are divided into the categories of sex, emotional, information, recreation, responsibilities and communication.

Knowing each other’s answers (correct guesses) in the communication questions category was one of the biggest contributors to relationship health scores in discouragement, communication, and fighting. This shows that communication really is the key to a healthy relationship.

Sexual health of a relationship was strongly predicted by couple’s knowing each other’s answers (guessing correctly) in the sex questions category.

“The study found that one of the most important factors for the sexual health of a relationship is simply how well people understand how their partner sexually,” Perron said.

His research was part of Insight’s 7-week fellowship program, which helps PhDs and Postdocs transition into the industry for careers in data science, data engineering, and artificial intelligence.

Besides demonstrating that technology can help people have better and healthier relationships, Perron’s study gives the Happy Couple team guidance and tools in how to optimize the app to always provide users with the right questions at the right time to help them build healthy, happy and long-lasting relationships.

Using the power of science for better relationships

Happy Couple, a relationship-building app, is coming closer to having the formula for healthy romantic relationships and utilizing the predictive power of relationship data to help couples have better connections. Over the past two years, Happy Couple app has been downloaded more than 600,000 times. Currently, there are 3,000 questions online, that make up a fun newlywed-style quiz game. Couples each answer 5 questions about themselves and 5 about their partner on a daily basis in order to get to know each other better.

The app also has more than 120,000,000 answers from these couples. With these tools, and the collaborative expertise of a data scientist and Happy Couple’s relationship therapist co-founder, this is just the first look at the predictive power of this data and what it can tell us about how to have healthy romantic relationships.

“Our next step will be for Happy Couple to regularly assess our couples with questions about how they are feeling in their relationship to add more context to their daily answers,” Robert said. “With this massive amount of relationship data, we’ll soon be able to start predicting when relationships are on the right path or not, and be even better at adapting our content to the right couples at the right time.

“With this massive amount of relationship data, we’ll soon be able to start predicting when relationships are on the right path or not, and be even better at adapting our content to the right couples at the right time.

Interested in transitioning to career in health data science? Find out more about the Insight Health Data Science Fellows Program in Boston and Silicon Valley, apply today, or sign up for program updates.

Data science for couples: Relationship health predicted by how well you know each other, not how… was originally published in Insight on Medium, where people are continuing the conversation by highlighting and responding to this story.

Derek Araujo was a Fellow at Insight Data Science during Fall 2017, and now works as a Data Scientist at Schireson Associates in New York. After earning an advanced degree in law, Derek started his career at an international law firm, and later served as legal council at a think tank.

However, Derek explains that he “missed using the quantitative side of his brain” in his law career and decided to go back to his undergraduate focus and long standing interest: Physics. He went on to earn an M.A. — and later a Ph.D — in Physics at Columbia University.

During graduate school, Derek “developed an interest in data analysis and learned to love the art of coding.” He learned about Insight Data Science through his graduate school colleagues, and believed the program could be a great fit. Data Science had become a burgeoning field in New York by that time, and he felt that a career in the field could be the perfect focus for his background and skills.

In an interview with SwitchUp, Derek talks more about how he got started in data science, his experience at Insight, and his advice for anyone considering a data science fellowship program like Insight Data Science.

You hold a Ph.D. in Physics, and an advanced degree in both Physics and Law. What made you decide to start a career in Data Science?
Data science was a natural career fit for me. Although I practiced law for several years, I earned my baccalaureate in physics. I ultimately left legal practice to pursue a Ph.D. in physics, partly because of my longstanding interest in the subject, but also because I missed using the quantitative side of my brain. During graduate school I developed an interest in data analysis and learned to love the art of coding. By the time I graduated, data science had become a burgeoning field. A vibrant professional community had matured in and around New York City, where I very much wanted to remain after graduating from Columbia. So the match was perfect.

“I ultimately left legal practice to pursue a Ph.D. in physics, partly because of my longstanding interest in the subject, but also because I missed using the quantitative side of my brain.”

How did you choose to attend Insight Data Science? What was your process to research Data Science programs?
I learned about Insight through word of mouth while in graduate school. Several Columbia physics Ph.D.s ended up at Insight, and spoke very highly of the program and its job placement record. The fact that the program is tuition free made it all the more appealing. I submitted my application just after graduating, and had planned on applying to Data Science bootcamps or going directly on the job market if I wasn’t admitted. Fortunately Insight accepted me soon after I interviewed with them.

What was the application process like? How did you prepare?
The application process was fairly streamlined. Candidates submit a written application and questionnaire describing their experience and interests. The Program Director selects a subset of applicants for thirty-minute Skype interviews. During the interviews the candidates are asked to talk about their academic research and review some code they have written. Admission decisions follow pretty quickly. In my case I was accepted within a day or two.

Prior to applying I devoted a good amount of time to self-studying machine learning algorithms, many of which I hadn’t encountered as a physics Ph.D. student. I’m certain that served me well. In retrospect, however, I may have over-prepared for Insight’s Skype interview. I spent the better part of a weekend completing a one-off project designed to showcase my machine learning and Python coding skills. My interviewer ultimately spent only ten minutes or so reviewing a small portion of my code. I suspect the interviewers use the code review to look for what would interest any employer: clean and readable code, but also oral presentation skills and an ability to explain technical concepts in an accessible way.

“I spent the better part of a weekend completing a one-off project designed to showcase my machine learning and Python coding skills.”

Tell me about the learning environment. What was the curriculum like?
There was no formal curriculum to speak of. Insight’s program is a full-time, seven week fellowship. During the first three weeks of the session each Fellow selects and completes a data science project he or she will present to employers. The remaining four weeks the Fellows help one another hone their presentations, prepare for technical interviews, and make their first presentations to employers. The learning environment is highly collaborative. Fellows come to Insight with varied skill sets and experience, and are expected to help one another acquire new skills rapidly throughout the program. In the end we ended up learning a great deal from each other.

“The learning environment is highly collaborative. Fellows come to Insight with varied skill sets and experience, and are expected to help one another acquire new skills rapidly throughout the program.”

What was your biggest challenge in the program?
My biggest challenge was finding the time to complete my project while also attending the numerous break-out sessions and employer presentations that were scheduled during the day. The first portion of the program was very intense. We had barely three weeks to collect and clean our data, test our models, and bring our projects to completion. That was a tall order, considering that half the day was spent in various meetings. Although this required some long nights of coding, the end result was satisfying. It also taught us to quickly identify which aspects of our projects had the highest return on investment, and which were expendable.

Insight Data Science boosts a diverse network of fellows. Has access to this community influenced your work as a Data Scientist?
It certainly influences my work at Schireson. Many of the data scientists I work with every day are Insight alums, and we work collaboratively, just as Insight Fellows do. I also make an effort to keep in touch with the Fellows from my Insight session and previous alums, who work in just about every area in the industry. I am still fairly early into my career as a data scientist, but I can easily imagine our benefiting from one another’s professional connections and technical expertise as our careers progress.

How did Insight Data Science help with your job search process?
Insight’s program directors played an instrumental role in helping me to navigate the job market. They facilitated introductions to prospective employers, helped the Fellows to manage their interview schedules, held mock interview sessions, and offered valuable feedback and advice throughout. The Insight Fellows themselves obviously do the hard work of completing the projects they present to employers. But the program directors served as an important catalyst by helping Fellows to identify and connect with companies that suited their interests and talents.

Now that you work as a Data Scientist at Schireson Associates, what is your day-to-day role like?
My role at Schireson has been satisfying on multiple levels. Schireson is a strategic data science consulting firm. My work involves a blend of coding and modeling on the one hand, and client-facing work on the other. Much of my time is devoted to the bread-and-butter work one would expect as a data scientist. But at the end of the day we need to derive actionable insights from that work and communicate them clearly to clients, many of whom do not have technical backgrounds. That inevitably requires translating technical jargon into plain English. But it also involves helping clients to approach their business needs from a data analytic perspective, or helping them to identify important questions they might not have thought to ask in the first place.

How do you use the skills learned at Insight Data Science in your new role?
The firm I work for draws upon a wide variety of models and techniques, both traditional and cutting edge, depending on the needs of a particular project. I had my first hands-on experience using some of these models while at Insight. Boiling down my Insight project’s results into a brief slide presentation and pitch was also a valuable experience. We are often required to do the same for clients.

What are your career goals going forward?
My overarching goal is to continue working in an environment where I am constantly learning and developing as a data scientist, while hopefully helping my colleagues to do the same. In addition to providing a large measure of personal satisfaction, it serves an important practical end. Data science is a rapidly evolving field, where stasis can lead quickly to obsolescence. Five years from now I expect to be using some techniques that are only now just maturing, or that haven’t yet been invented. If I remain sufficiently agile to evolve in tandem with my chosen field, I will be happy.

What advice do you have for people interested in attending a Data Science Fellowship?
I can only speak personally to my experience at Insight, which was uniformly positive. Regardless of which program interests you, it helps to talk to current and former attendees about their experiences, their programs’ strengths and weaknesses, and what preparation they found to be most helpful. In general, be prepared to work hard, and to learn quickly. Most of all, be open to learning from your colleagues and offering them help in return when they need it. This collaborative approach makes learning much more efficient, while also helping to identify any gaps in your own understanding.

“…be prepared to work hard, and to learn quickly. Most of all, be open to learning from your colleagues and offering them help in return”

Interested in transitioning to a career in data science? Find out more about the Insight Data Science Fellows Program in Silicon Valley, New York, Boston and Seattle (deadline June 25, 2018), apply today, or sign up for program updates.

From Law to Data Science was originally published in Insight on Medium, where people are continuing the conversation by highlighting and responding to this story.

We have witnessed the growing need in a front-row seat through building the Insight data community in Boston since 2015. We are particularly excited about the expansion to Boston, a city with an unique blend of academia and industry to make innovations in technology. This year, Boston ranked as the #1 biotech hub, with $2.9 billion venture funding in the past year, giving professional engineers immense opportunities to make a positive impact through technology. Many top-tier data teams are drawn to Boston thanks to the burgeoning supply of talented students.

Still, the demand for hiring data engineers has increased dramatically due to a rapid growth of data in volume and complexity. Companies are looking for experienced software engineers to build scalable, distributed data architecture to process massive real-world data. However, there is an increasing number of specialized open-source technologies in the data engineering ecosystem. It is extremely challenging to find candidates with hands-on experience in big data engines (e.g. Hadoop, Spark, and Cassandra) as well as an ability to collaborate with other team members such as data scientists or product managers. That’s where Insight comes in.

The Insight Data Engineering Fellows Program is a 7 week, full-time, professional training program. It enables experienced engineers and programmers to apply their existing software development skills by building efficient and scalable data platforms with distributed systems. Throughout the fellowship, Fellows receive mentorship from practicing data engineers at companies such as Amazon, Microsoft, Tamr, Wayfair, Boston Consulting Group, Boston Health Economics, MassMutual, Kayak, iRobot, TempAlert, among others. Several of these companies will be looking to hire Fellows for positions in Boston.

In addition to learning data engineering tools and building projects under the mentorship of Boston-based engineers, Data Engineering Fellows will have opportunities to gain practical machine learning experience through collaboration with our data science Fellows. Moreover, they will join an active network of over 1,000 Insight alumni now working as data scientists and engineers at over 350 companies across the US.

We are looking forward to helping engineers and programmers transition into data engineering in the Greater Boston Area. Applications for the April 2018 Boston session are now open, as well as for upcoming sessions in Silicon Valley and New York City — Learn more and apply at the Insight Data Engineering website.

Interested in transitioning to a career in data? Find out more about the Insight’s Data Science, Data Engineering, Health Data, and Artificial Intelligence Fellows Programs in New York, Boston, Seattle and Silicon Valley, apply today, or sign up for program updates.

Insight Data Engineering Fellows Program Expands to Boston was originally published in Insight on Medium, where people are continuing the conversation by highlighting and responding to this story.

Insight Data Science is honored to have been featured this week in the New York Times article “Where the STEM Jobs Are (and Where They Aren’t)”.

Interviews from Jake Klamka (Insight Founder & CEO) and two alumni, Carlos Faham and Anasuya Das, shed light on how the rapidly growing field of data science is absorbing graduates from STEM domains.

Jake Klamka, a physicist who founded the [Insight] program, kept hearing from Silicon Valley executives that they had considered hiring traditional scientists, but converting them to technologists seemed time-consuming and risky. So Mr. Klamka decided he would start a company to provide scientists a smoother pathway into the tech industry.

Carlos and Anasuya describe their own career moves from academic science into data science careers…

Data science is distinctly different from neuroscience, Dr. Das said, but some of the tools she employs, like a machine-learning technique called artificial neural networks, do take their inspiration from the brain. Her experience points to the larger trend that digital technologies like data science and artificial intelligence are increasingly being used in nearly every discipline.

… and some of the challenges they faced in making that transition.

“It was like hitting a wall running at full speed, really humbling,” [Carlos] recalled. “Interviewing is a muscle and you have to exercise it again and again,” he said. After the [Insight] program, he received six job offers. He accepted the offer from LinkedIn. The range of data-intensive detective work, he said, is “extremely rich” and “it moves so much faster than my previous world.”

Read the full article here.

Interested in transitioning to a career in data? Find out more about the Insight’s Data Science, Data Engineering, Health Data, and Artificial Intelligence Fellows Programs in New York, Boston, Seattle and Silicon Valley, apply today, or sign up for program updates.

New York Times features interviews with Insight founder and two alumni was originally published in Insight on Medium, where people are continuing the conversation by highlighting and responding to this story.

Insight Data Science alumni Anasuya Das and Amelia Taylor were recently featured in a Nature story about the transition of academic scientists into data science career tracks.

Anasuya came to Insight in the winter of 2015, following a PhD and postdoc in Neuroscience from the University of Rochester Medical Center and New York University, respectively. She is currently a senior data scientist at the Memorial Sloan Kettering Cancer Center in New York.

During Amelia’s time as an Insight Fellow in the summer of 2016, Amelia Taylor consulted for the water utility company Valor Water Analytics to detect cases of water meter tampering. Prior to Insight, Amelia was a tenured professor of Mathematics at Colorado College in Colorado Springs. She currently works as a data scientist at Zymergen in Seattle.

In the feature, Amelia and Anasuya speak about the challenges of understanding the data science field, and the range of competencies that are required on the job.

“Data science can look very different at different places. There are so many companies out there, it’s hard to know which ones to look at.” — Amelia

“My days vary widely, and range from doing pure software engineering to meeting with physicians about the products we’re building.” — Anasuya

The story also discusses Anasuya’s introduction to data science through her research career…

As part of her neuroscience training at New York University and the nearby University of Rochester, Anasuya Das, a former Insight fellow, had to learn the coding language C++ to build software to help individuals recovering from strokes to practise visual learning using their home computers. Das also took a couple of computational-neuroscience courses that helped to spark her interest in data science as a full-time career.

… as well as the benefit to Amelia of Insight’s industry-based approach to learning.

The Insight fellowship was invaluable in teaching [Amelia] the skills needed to land her current job. Among other proficiencies, that training taught her to think beyond data analysis to the practical applications of the finished product. “The business-oriented thinking at Insight was very helpful,” she says. She has observed that PhD-level scientists who land data-science jobs in industry tend to struggle with the transition unless they’ve already had first-hand industry experience. “I had a very fast start with my company because of my ability to think about products”

You can find the full article here to learn more about the surge of data science as a career choice among scientists and researchers.

Interested in transitioning to a career in data science? Find out more about the Insight Data Science Fellows Program in Silicon Valley, New York, Boston and Seattle, apply today, or sign up for program updates.

Nature story on data science careers features Insight alumni was originally published in Insight on Medium, where people are continuing the conversation by highlighting and responding to this story.