Mickey Petersen Email & Phone Number

London, England, United Kingdom

Zürich Area, Switzerland

Brought in from London to help architect and lead the development of the Consolidated Capital Model – a Monte Carlo-driven model built from scratch and scaled to terabyte-ram-sized analyses when I left – for a front office team of actuaries and quants. I set the technical direction for the model – mostly developed by actuaries, who were beginners at Python – and ensured the resulting model met the practises expected of a modern software project: turn-key deployment from development to production and fully dockerised; complete BDD and Unit Test coverage; and a strong focus on ensuring the team wrote maintainable Python code. I trained the team in Python and helped ensure the model performance met the demands of the business. Further to the model, I also built the web application (Flask, Postgres, Celery, Docker) for running and managing the lifecycle of model runs.I acted as the primary focal point between the Business and the IT department, and ensured the team had the requisite servers and infrastructure needed to function once we went live.Following on from a successful, and on-time, release of the model, I started migrating the model to Azure, to better benefit from the scale-out facilities available in the cloud.When I joined the department did 30-40 model runs a quarter using a legacy COTS tool. By the time I left, we did 20,000 model runs in a year alone. All of it elastically scaled and right sized to the exact requirements of the model and the parameters it was given. Along the way, we added on another five actuarial models, and I built a sophisticated graph tool to analyse data inputs and outputs and optimise the order in which the models were run; when to delete output data to save space; and how to retrieve data from dozens of different data pipelines.

London, United Kingdom

London, United Kingdom

Hired to turn around a large media company's data ingest and transformation systems that were using a litany of technologies to move data around.As architect, I ran a team of developers and devops and ensured best practices were followed and put in place a plan to migrate crontab-based ETL and data ingest shell scripts to a modern, maintainable Apache Airflow architecture powered by Kubernetes.

Amsterdam Area, Netherlands

London

Short-term contract to develop a natural language processing tool that scrapes Twitter's live feed and determines what people are talking about.

London

Brought in to optimise a biotech application for use with CRISPR/Cas9. I spent most of my time optimising PostgreSQL queries and the design of a very large database so as to make better use of available resources. I also re-architected the Python code so it would scale better horizontally and vertically using Celery and RabbitMQ to support large-scale cloud provisioning.I also formalised their configuration/deployment management using Salt and maintained their Python/Pyramid web application.

London

Short-term contract to refactor a SPA procurement application written in Python & Flask using SQLAlchemy and PostgreSQL.

London

Django, Python, Javascript, PostgreSQL.

London

Architected and developed a cloud-based Software-as-a-Service monitoring system for servers and back-end services. Our servers would receive telemetry from agents hosted on our customers’ servers, then collect the data for use in “Big Data” analytics, powered by Cassandra and Pig for the storage and analytics; Hadoop for data processing; and with the infrastructure written in Python with Flask, Javascript (JQuery), PostgreSQL and SQLAlchemy.

London, United Kingdom

Part of a larger team involved with gathering, scraping, parsing and warehousing a large amount of economic research data, retrieved from a variety of websites in an untold number of formats; this data – all of it economic time series – is then sanitised and put into Oracle for use by traders, analysts and economists. My work involves building complex web scrapers; bespoke ETL tools to handle the extracted data; and general Python and Oracle expertise to a team of analysts and developers.

London, United Kingdom

Short-term contract to help develop and integrate a cloud-based infrastructure capable of handling RFID card “interactions” and large amounts of media content, triggered by visitors “interacting” with touchscreen stands at auto shows around the world, then collect and analyse the data for use by a large automaker’s marketing department to extract customer-, and market trends.

Hired to work on a release framework written in Python used worldwide for all of GBM’s internal applications, my role was to maintain and extend the existing framework and provide Python expertise and knowledge to the release management team, and assist with the transition of existing software deployment processes to the new framework.

Hired as a Python expert to review and standardise Python practices and develop bespoke backend solutions to help Glencore transition from an old Treasury system to a new one.As part of my role, I redesigned the internal source control and release management procedures across all development teams to improve release quality and minimise overhead; this mostly took the form of training and creating release procedures suitable for small development teams using Subversion and Hudson.

I was one of four principal developers charged with creating a financial model for allocation of passengers and revenue to trains.I greatly improved the development cycle by organising the development practices such as introducing a wiki to capture important knowledge about our product; develop automated UI testing to guard against regressions; introduced continuous integration; developed performance metrics to ensure our model met performance targets and contractual obligations between releases.

I was the developer in charge of maintaining their in-house CRM system, written in Delphi. The tool was ciritical to the operation of the company, as it tracked all their sales leads, issue tracking and technical support tickets.

I formalised their ad-hoc software processes and introduced several optimisations by automating manual tasks using Python. Several manual data entry processes were eliminated resulting in man-years' worth of savings.