# Mohamed Bana
My main interests are in Rust. And also Golang/Go in the area of the Language Server protocol, see my contributions to https://github.com/saibing/bingo in my fork https://github.com/banaio/bingo. The Language Server,
bingo in this case, is used by
vscode-go to provide Golang language support in the IDE. I am passionate about (developer) tooling.
I am a well-rounded Software Engineer, and I understand what it takes to release a product having worked on the back-end, front-end, testing and finally the deployment aspect of several products.
I am looking for a role involving web technologies hosted on a cloud-based backend. I’ve worked on all the tiers of a software product, so I can appreciate the different concerns expressed at each level. I feel the latest trend in web technologies; quick prototyping, large selection of libraries, ease of programming and its cross-platform support is where the future is heading. This, distributed systems and big data algorithms are where I am focusing my current efforts—all of which are equally interesting to me.
- 2008-2009 — MSc Software Systems Engineering (Attended), University College London, and Trading & Financial Market Structure module, London Business School.
- 2005-2008 — BSc Computer Science (2.1), City, University of London.
# Work Experience
# Full Stack Node.js Developer (contractor), Open Banking Limited, St Katharine's & Wapping, London - 08/05/2018–Present
Working as a full stack developer at Open Banking on a tool that will validate a bank’s implementation of the OpenBanking API spec, see:
# Senior Engineer (contractor), Ninety Percent Of Everything Limited, Marble Arch, London - 01/02/2018–20/04/2018
Worked at startup specialising in software that runs on ship on two projects that were heavily Go-based.
Document storage and retrieval to be used by others services, so it's a core service. The core of the service was written in Go and exposed using gRPC and http using go gorilla/mux. Both write and retrieve supported arbitrarily large files which was achieved using gRPC unidirectional streams. The underlying store was MongoDB's GridsFS which I interfaced with using the Go driver mgo.
Currently working on this. The service is structured very similar to the preceding in that the underlying service is exposed using gRPC but the top-level interface is done using GraphQL. I wrote the GraphQL server in go using graphql-go. The underlying store is in Postgres and the library I used to interact with it is GORM.
Tech: Go, Golang, gRPC, Protocol Buffers, MongoDB, GraphQL, go gorilla/mux, Docker, Docker Compose, Kubernetes, NodeJS, Jest, Concourse CI, Postgres.
# Full Stack Developer (contractor), Root Capital LLP, London Bridge, London - 09/10/2017–24/12/2017
Worked as a full stack Node.js developer on the Minds for Life application, mainly on the forum.
- Single Page Application (SPA) targeting mobile platforms.
ES56using most of the latest ES56 features;
- Serveless and hosted on Amazon S3 as static assets, with Amazon CloudFront as the CDN.
- NodeJS server written in
ES6, like the frontend.
- MySQL as the datastore, using the Knex.js library.
- Packaged as a set of
CI, Devops and Infrastructure:
- Services were packaged as containers. Used
docker-composeto start them.
- Builds managed by Semaphore CI and Wercker.
# Full Stack Node Developer (contractor), Lloyds Banking Group PLC, London Bridge, London - 20/03/2017–22/05/2017
- Loopback for server-side of the code.
- ES5/6-based code base.
- Splunk and sending logs via (https://en.wikipedia.org/wiki/Syslog) a LogDrain service available on Bluemix.
- Gerrit for managing code. CI:
- Jenkins: Configuring, managing and installing.
- Jenkins 2: Same as previous plus writing pipeline scripts.
# Senior Front-End Engineer, Synthace Ltd. King's Cross, London - 13/04/2016–04/11/2016
Did a fair amount of architectural UI work:
- JWT-based authentication: Implemented most, if not all, of the authentication related UI features. Polymer didn’t have an authentication module as it’s fairly new requiring me to reimplement this feature.
- API interactions: I introduced Swagger JS and did the conversion from plain XHR to Promises, and ensured API was in-sync with the state of the authentication.
- Updates via the Web Socket for notifications and async task updates: STOMP Over WebSocket.
- Bootstrapped the testing using Web Component Tester.
- Deciding on the build, test and hosting strategy, e.g., hosting our own CDN using Azure.
- Performance: 1) pushed to have HTTP/2 enabled, and prototyped, on our custom server written in Go, 2) implemented lazy-loading of our Web Components which are included using HTML Imports, 3) Significantly improved UI build scripts; went from a somewhat un-deterministic build to one that almost always runs.
- Introduced ES6 to the code-base, and moving to defining Polymer elements using ES6 classes.
- Misc: libraries/utils to ease UI development.
We deploy Docker images to our Kubernetes cluster running in Azure using:
- Docker: Fairly comfortable using this.
- Kubernetes: I've done deployments of dev branches, so I understand the deployment model, navigating the Kubernetes dashboard and crude command line interactions, e.g., port-forwarding of the service the pod is running from the cluster to the local machine.
I’m finishing up on adding support for our language Antha to Monaco Editor, the editor that powers Visual Studio Code.
Since this is a startup I have done a fair amount of work and I have been given a fair share of responsibility, more so than any of my prior roles.
# Developer, IG Index Ltd. Cannon Bridge, London - 09-2014–18/02/2016
- Charts: Assisted in the conversion of the Adobe Flex Real Time Charts to an SVG-based version. Tech: d3 and tested like above.
Due to the nature of the work, I cannot disclose too much detail.
# Software Developer, ITRS (International Trading Room Software) Group Ltd. Moorgate, London - 02-2010–09/2014
UI for the next generation of the product which is built around, loosely speaking, a real-time distributed analytics store. The aim is for the old system to stream data to the new system so we can provide all the great visualizations available in the HTML ecosystem, which was not achievable in a reasonable amount of time in Swing.
The code is entirely modularized using RequireJS so we can test each viewmodel without creating a view, we then test the entire UI (end-to-end tests) using WebDriverJS.
Some Java/C coding required to write NodeJS bindings to interact with the store. We evaluated several frameworks, Angular, Batman and KnockoutJS, by writing prototype applications that connected to our backend for a period of roughly 4-6 months before we chose to settle on Knockout.js.
# Java (UI) Developer - 07/2012–12/2013
Worked with two developers and one QA member on a new Swing UI, ACLite, that uses our new streaming-based API to access Geneos data, for more info. see https://resources.itrsgroup.com/OpenAccess. We access data from a fault-tolerant Cluster that is built on a set of distributed nodes. The system we coded against is somewhat similar to the Amazon distributed key/value store, DynamoDB, except with support for streaming, so I am familiar with dealing with distributed systems.
Tech: Java, Maven, Swing, Eclipse, Jenkins, Git, Vagrant.
# C++ Developer - 02/2010–06/2012
Spent one year with Run The Business (RTB) team, a team set-up to fix critical bugs that Customers encounter. A very challenging role which requires all-around product knowledge, good debugging skills and being able to liaise with our Support staff in dealing with the Customers. Prior to this I was one of three developers working in the Transactions and Latency Monitoring team (part of the backend team) doing core C++. We wrote and maintained the following plug-ins that are part of the Geneos suite:
- FIX-Analyser: Monitors FIX (protocol) messages.
- Feed Latency Monitor: Monitors feeds and calculates latency of instruments and fields across the monitored feeds.
- Message Tracker: Tracks, generally, FIX messages across several checkpoints.
- Market Data Reliability: See below. And bespoke plug-ins written for specific firms.
We also maintained several non-finance specific plug-ins. I ported another bespoke plug-in called Price Latency Monitor (provides latency figures for bonds) to MS VC++ when I worked on this team. Projects:
- I converted the Windows version of the entire product suite from Visual Studio 2005 to 2010.
- I wrote the Market Data Reliability plug-in. This plug-in connects to the Patsystem's Trading API (www.patsystems.com) to monitor commodity prices, using their C API, to determine if prices are 'stale'.
- I ported a significant part of our product to Solaris x86-64 (64-bit non-sparc architecture).
Tech: C++, STL, Boost, Visual Studio, Linux/Unix, GDB, DBX, Make, Configure, XML, XPath, CPPUnit.
# Software Developer Intern, then Tester, Thomson Reuters. Canary Wharf, London - 05-2009–11-2009
# C# Developer
One of four developers working on a search and navigation interface to Global Product Search. Full life-cycle of development; requirements engineering, analysis and design to implementation, testing and deployment.
Tech: C#, Silverlight 3.0, MS SQL Server 2005, LINQ, Web Services (WCF), XML and Visual Studio 2008. I handled deployment using CruiseControl.NET.
User Acceptance Testing of the latest release of Thomson Reuter's 3000 Extra, then called UTAH, now called Eikon. UTAH combines the data from Thomson and Reuters. My primary responsibilities were to validate the end product against pre-defined requirements/workflows. 1. Worked on Thomson Reuters project UTAH as part of a large team. 2. Tasks included testing, observing, documenting software bugs, issues and errors before final release of Utah. 3. Testing was done over multiple iterations.
# Additional Info
Full UK driving licence.
- English and Swahili: Native.
- Arabic: Basic to intermediate.
I’ve played with Scala when it was in its infancy. I wrote an Eclipse plug-in, https://github.com/mbana/MOODMetricsPlug-in, which calculates the set of metrics called MOOD (Paper: An Evaluation of the MOOD Set of Object-Oriented Software Metrics) on Java soure code. The plug-in works by using Eclipse’s internal compiler API to get an AST of the Java code and calculating the metrics.