Enrico Carniani

A Flask-based back-end app written in Python3 that can track cryptocurrencies and invest automatically via exchange APIs of Crypto.com or Binance.com.

The app's UI allows the user to create multiple token models based on hundreds of features. Some statistical tools automatically assist the model creation, favoring the best parameters and synthesizing new compound features.

Once the model is trained, it can be further enhanced by an assisted hyper-parameter tuning. Its performance in time is also monitored to signal the degradation and necessary retraining.

The machine learning part is written using Python's scikit-learn framework and some other libraries, such as XGBoost. Models can support different classification algorithms, including time-based ones.

The app, supporting multiple operation queues to exploit multiprocessors and foster interactivity, runs continuously in the background, saving crypto-market values to a MySQL MariaDB instance.

Model instances are then run against the latest values to trigger actions such as selling or buying new cryptocurrencies. The app can handle an arbitrary number of currencies/tokens connected to the crypto.com exchange.

A React Native app for both iOS and Android aiming at helping people to reach and maintain their mental fitness. The app serves media content, such as video, audio, images, and texts, divided into classes of different levels. The routine is gamified and users assign badges depending on their activity.

I also wrote the back end in LoopBack 4 on top of Node.js, available as RESTful APIs with JSON. The authentication and authorization are performed through JWT with Auth0 as the provider. The content is hosted by AWS S3, distributed via CloudFront CDN, and the server has been turned serverless and boxed into an AWS Lambda instance.

As middleware, I also wrote a simple CMS in Python 3 as an agent able to upload, transcode, and distribute content for the app to AWS.

A multiuser platform written in CodeIgniter4 (PHP7), ES6, CSS3, and HTML5 for NewRadio s.r.l., a company specializing in services for the radio industry and managing multiple web-radio players per account.

The player's skin can be picked among many standard ones, or they can choose one specifically designed for the client, as the player is pure HTML5 with a well-defined structure able to host flexible CSS themes. Special CSS effects like animations are available for stunning effects.

The current track title playing is supported via API services. The back end acts as a proxy and a cache for the different available sources, including Icecast, SHOUTcast, and others.

The web client player written in ES6 is obscured and encrypted through an API license. Almost all web audio decoders are supported through a fallback mechanism that ultimately launches a Flash player in the background. Play5 aims to play music anywhere.

Track art (covers) are also supported and loaded automatically by the client from Apple iTunes services or custom software specialized URLs. This is one of the many features that the company owning the player can switch off.

Developed a PHP-based site offering coupons to online casino players. Explicitly not written in React, but still exploiting full Ajax support and using ES6 on the client side.

I wrote it all from scratch, including the CMS feature, which is currently not visible, automatic language translation via Google, and total coupon management with user events such as views and clicks.

Registered users are also managed so they can be ranked and offered unique gifts. The service also provides an API so that an iOS app, now removed from the store, can also offer an excellent experience for this service.

Wrote an extension for most common browsers, such as Chrome, to show coupons when the user enters sites the platform has coupons for.

A project to manage the network of the University of Bologna, Italy, made on behalf of Laboratori Marconi SpA. It is a suite comprising different apps, such as:

• ALMAVIEW NG web interface, a web-based application written in Zope 2, Python 2, and a multi-user role-based UI. Its objective is to maintain the infrastructure of the campus network. It acts as a centralized interface of registered devices queried via the Simple Network Management Protocol
(SNMP). Devices are organized by faculty, department, and area. The tools associate human managers with device groups. It is also able to create configurable PDF reports.

• Netgear NG with C++ Probe, a super-fast real-time network analyzer capable of sampling information on behalf of AlmaviewNG. Samples and meaningful stats are sent to AlmaviewNG in real time. It acts as a vampire on the wire and checks for suspicious warms or illicit activity such as copyright infringements and security health checks.

• Python 2 agent, an agent that performs other network tests not requiring real-time analysis capability. The agent analyzes and measures the network characteristics and fragilities via SNMP and other protocols using other command-line tools such as Nmap.

An iOS and Android app used to gather electromagnetic pollution via a Bluetooth device, a patent from the same company, formerly named Drayson Technologies, in London.

I designed and implemented an events processing pipeline in C++. It can process data from all onboard phone sensors, plus the external Bluetooth device.

The pipeline input was a chain of filters to normalize and treat data and had an embedded, downloadable model to predict user behaviors and pollution trends. Classifications and other data were then sent to the app for further processing and displaying.

One of the main peculiarities was its ability to be set up and configured remotely; the model had versioning and was periodically regenerated on the cloud side.

The framework was able to run on iOS and Android. I integrated both platforms by using respectively objective-C++ and JNI as glue.