CROWDX is a state-of-the-art crowdfunding platform built with Web3 architecture including blockchain, smart contracts, programmable patronage badges, and distributed file storage which enables new forms of patronage in the digital age, allowing creatives to maximize value creation with their community in a manner not existent on incumbent platforms

After being contracted through my company 135B to develop the architecture and smart contracts of the REFD platform, I continue to provide technical guidance to the team while they continue development.

In September 2015 I've won the London Fintech Week blockchain hackathon and soon after I joined Thomson Reuters as a contractor blockchain developer. My responsibilities are around defining the architecture of distributed applications, building private Ethereum network infrastructure, coding smart contracts and the additional JS layers needed to interact with them. One project where I had a significant contribution, Blockone ID is now launched in public beta. It maps digital identity to on-chain identifiers and also offers a hosted wallet for ease of use. I demoed the project during a hackathon organised in September 2016 by Thomson Reuters and later to the London Ethereum meetup.Another part of my work here is to research other blockchain fabrics such as Eris, Tendermint, Hyperledger, Multichain.

My primary focus and project ownership in the team of 5 people was disassembling and reverse engineering of multiple poker client applications for extracting real-time local data from poker tables. With my work, the performance of the process injection module was improved by one order of magnitude. As tools, I used IDA for disassembly, C++ for writing code, Frida as injection library and Python for automation. I also worked on architecting and implementing the update automation platform where I employed Jenkins as job automation tool and Python and Bash scripting as glue code. I also ported the existing QT client codebase to Mac OSX. By working fast and proficiently I helped streamline faster the roadmap for the Mac client.

I was part of the Robot Detection team whose target purpose was to identify and mitigate robot generated web traffic by analysing petabytes of data coming from amazon.com, its regional locales and affiliated websites. I was part of the project from its infancy and had important contributions in architecting the MapReduce Hadoop-based hourly jobs. Given the commercial magnitude and technical scale of the problem to solve, performance was crucial. With algorithmic changes in key bottleneck points I managed to improve the running time of the jobs by a factor of 5. In order to develop the robot pattern detection models, I leveraged the existing Oracle-based data-warehouse and business intelligence solution to analyse relevant time-series data slices. As a side-project at first towards approved production use, I proposed and developed a machine-learning system for tracking new robot patterns. For this, I used Random-forests and Bayesian algorithms from Apache Mahout framework. Java and Scala were the main programming languages used, together with: Python for scripting, *nix CLI tools, Oracle SQL, d3js data visualization, Junit and Mockito for testing.

I started my career early, while still in the University by joining Bitdefender as an antivirus researcher. This was my formative period, where I learned and worked by doing everything from analysing low level assembly code to writing a new compiler and virtual machine for an internal language similar to Java. I had important contributions in multiple important projects and was constantly promoted with outstanding marks. To enumerate a few:1) I wrote a Windows Kernel anti-rootkit driver in C that was used as the backing tool for the malware cleaning routines. This project spawned from my bachelor’s degree university graduation project which was graded with 100% credits.2) Another important project that spawned from my academic background which was the core subject of my Ph.D. thesis was a large scale Machine Learning system that contributed to the main business-case of the anti-malware product – detecting malware files. Because scale and performance were key, we wrote a multi-threaded framework from scratch in C++, including the GPU/OpenCL modules. Most of the work was published as academic articles.3) I managed two middle sized teams. The team that handles the large scale crawling system responsible for populating the clean-set of executable files and the team that handles disinfection on the Windows platform, part of Bitdefender core technology.Keywords: C/C++, X86 ASM, Python, SQL, algorithms, data structures, compiler design, kernel driver, OOP, Java