Over the past 10 years, the volume of global fraud losses for online payments has increased at an average rate of 10% each year. As of 2014, the global losses due to fraud are expected to reach the astounding amount of $14bn, which is equivalent of 140,000 jobs.
In what follows, we first describe the two factors that explain the exponential growth of online fraud over the past ten years, and then we review the fraud losses faced by an average US company, hypothesising on the factors that increase and that decrease fraud rates for companies.
Nowadays online payment processing is so efficient that we never question how it works behind the scene. We go to Amazon and we buy best-selling books, trendy clothes, and even fresh food. Or we go to Expedia and we order an airline ticket worth $1,000. We receive an email confirmation that says that the ticket is issued, but nothing else.
In what follows, we will review the business entities that make online payment processing possible. It starts with the merchant website, and it continues with payment gateways and payment processors. It involves credit card unions and it involves three
In this post I review the sales funnel process, its engineering and its recent changes. The sales funnel is the term used to describe the selection process that occurs when converting potential customers into buyers. At any time a potential customer who enters the sales funnel is in one of three states. It is a lead, a prospect, or a buyer.
To become a lead a potential customer has to enter in our ‘radar’. To become a prospect a lead needs to be aware of our product. And to become a buyer a prospect has to realise a purchase.
Recently I worked at BuildingConnected.com, a social network for construction companies that use the platform as a means to bid for construction packages and to exchange between contractors and sub-contractors; as of May 2014, the user base starts expanding beyond California’s boundaries with more than 2,000 users. In need of support to develop new functionalities and to go with the growth of the platform, Jesse Pedersen, CTO of BuildingConnected.com, reached out to Hack Reactor three weeks ago.
In this post I do the making of foodbot.io, an Angular Google maps web-app collecting 150,000+ free food and drink events in Bay area, which I co-developed with Rob Graeber and Abdelatif Sebbane as part of Hack Reactor. First I come back on our tech stack, on the why of Angular Google maps, and on our top-5 features. Second I review the how to of Angular Google maps. And finally I wrap up with the main challenges faced when developing the front-end.
From eight years of research to the end of the impostor syndrome and from trading airline tickets to joining Hack Reactor. In what follows I sum up the path that led me to Hack Reactor.
Ten years ago, after my electrical engineering studies I started a phd because this was my dream. Then I landed an awesome postdoc at UCLA. This was another of my dreams too. Then I did a postdoc in statistical genetics that allowed me to lift up my mathematical skills and to rejoin my long time interest for modelling. Finally, I pitched my postdoc to an Harvard colleague’s research group. That’s exactly when I understood that I was no impostor.
That’s exactly when I got the confidence that I could tackle top-grade analytics problems. That’s exactly when I started to fly on my own...
In research I learned to value clarity, conciseness, and data-driven insights. Surrounded by bright scientists, I also learned the value of their time because it takes months to get university professors seat around the table to review your last findings...
After these eight awesome years in research I moved to a major online travel agent to prevent credit card fraud with statistical modelling. There within months I contributed a business impact in sales of $MM+ and cut manual verification workload by 30%. If I needed additional confidence in my capacity to do things, there it was...
Yet in spite of my dedication to research and to my significant contribution in industry, wherever I would go I would face exactly the same resistance: a resistance against change and notably to the change brought by web technologies that I made used of. Back in 1997 I did my first website. In 2000 I wrote my first search engine and since then I work on Unix (i.e. 14 years). Back in 2006 I used an in-memory queue to dispatch jobs to 150+ two-processors workers while benchmarking Support Vector Machines for my phd.
That’s when the Silicon Valley, San Francisco, and Hack Reactor come in. Here in San Francisco I feel like I am at the right place. Finally... At Hack Reactor I am surrounded by alumni totally dedicated to the study of software engineering. We are mentored by passionate teachers who line up hours because they just love what they do.
And we commit to positive thinking by lifting people’s self-esteem, by leaving aside negative perception and by insisting on the "yes" it is possible to be a full-stack software engineer within the three months of the bootcamp immersive. In short, being here feels totally awesome. People embrace the change and are willing to take on feedbacks to improve themselves and evolve; what a relief…
Still, now and then people ask me why I joined Hack Reactor with a phd in machine learning and postdocs in human genetics?
The answer is simple. San Francisco is the Harvard / MIT of the tech-world, I have always been a techie, and being here feels home. Besides, this is really where my know-how and my drive meet "the least resistance". This matters a great deal as it is fine to feel resistance when you are a teenager or a young professional but now I am a battle-hardened expert with eight-years of experience designing top-grade predictive models. So, no more detours. Just the Silicon Valley, San Francisco, and Hack Reactor.
Thanks. Justin Yek for discussing an early draft.
In this post I come back on an unexpected perspective on complexity from the Fibonacci series. First I recall who Fibonacci is, and what the Fibonacci series is. Second I review the algorithmic complexity of a top-down recursive calculation of the Fibonacci series. And third I present the benefit of reversing the problem, and of solving it from the bottom up.
In this post I come back on an unexpected finding when working on a toy algorithmic problem, bottom-up recursive n queens misses solutions. So in the following, first I reformulate the n queens problem. Second I describe a recursive n queens. And third I describe the new, permutation-based, non-recursive solution, and why recursive n queens failed.
In what follows I explore software scale up: why it matters, what we aim for, and an how to. First I set the context of software scale up. Of how come it is a topic of interest for me now. And of the general programming concepts that are used to support software scale up. Second I set the two objectives that we are after when aiming for software scale up. And third I explore the how to of making (web) software scale up.
Genetic research on auto-immune disorders suggests that haplotype HLA-DRB1/DQB1 is the highest risk locus in the Major Histocompatibility Complex.
One of my postdoc projects focused on analysing sequencing based HLA-genotypes of a dutch Rheumatoid Arthritis cohort. As the genetic data was a mix of allelic information at two and four digit resolution, my first task focused on imputing the higher resolution (four-digit) allelic information by Expectation Maximisation while assuming Hardy Weinberg Equilibrium.
Autism Spectrum Disorders (ASDs) is a group of developmental disabilities, including Autism,Asperger syndrom and Pervasive Developmental Disorders that can cause significant social, communication and behavioral challenges. People with ASDs handle information in their brain differently than other people [CDC].
Level of severity varies between the different subtypes of ASD in terms of social interaction, eye contact/joint attention, absence of response to their name or unusual movements. Being interested by Autism, I explored a bit the Internet for websites and video documentaries about autism. And, because of the high prevalence of autism in the population (in the US at least) as well as the “popularity” of autism (through movies, shows and famous autist people), there is abundance in resources covering the subject. In this post, I refer to a selection of links and video material that interested me particularly.
The scene in which epigenetics [1-4] is at play magnetises me, so I searched for epigenetics documentaries. In this post I relate environmental effect and genetics, position epigenetics in this context, and I link to three epigenetics documentaries, which I found amazing.
In this post I cover how to remove redundant information to speed up reading timetables. In the following, first I go back on the actual time tables. Second I define a way to compare an hypothetical time needed to read from A to Z a timetable. And finally I compare that time between the actual timetables and the proposal.
A few years back, as I was visiting my former colleagues of the LIAAD in Porto. I took the brand new metro do Porto. And being new to the metro system, I liked to figure out a few things. How frequently was each station deserved. How late and how early I could use the metro -to catch early airplane-. What were the routes and connections. What was the price, in which zones, etc.
Last July, Karin van Haren, program coordinator at NBIC, asked me whether I would accept to be interviewed on R SDisc. She liked to present SDisc in the Hands-on section of Interface, a bi-annual magazine published by NBIC and describing the developments in the field of bioinformatics and especially the research results from the BioRange sub-program of NBIC.
2014 © Fabrice Colas. All Rights Reserved.