UQ Summer Semester 2017
Mathematics & Physics Industry Project
(including information for SCIE3255)

Welcome to the School of Mathematics and Physics (SMP), Summer Industry Experience Project. This page is intended for students wishing to take part in the summer industry experience, either for credit (via SCIE3255) or not for credit. The program is organized by Murray Kane, Yoni Nazarathy, Tom Stace and Thomas Taimre. If you wish to participate for credit, consider the for credit section below. Industry partners - see industry engagement page.

• The program is intended for Bachelors, Honours and Masters students in the School of Mathematics and Physics. Bachelor students, may opt to participate in the program for credit (subject to a successful interview process).
• The program runs between November 2017 and February 2018. During this time, Mathematics, Physics and Statistics students can participate in paid internships for a duration of 6 to 8 weeks.
• Student participation in the program is subject to a successful interview process, facilitated by SMP and carried out by industry.
• For further details, consult the frequently asked questions section below and if needed contact the program staff (Murray for administrative question and Yoni for grading and project questions).
• You may also find further links dealing with careers, employability and general tips for life in the resources section below.

---

Applying

All eligible students are encouraged to apply. An application consists of an e-mail with an attached CV, an attached grade transcript and one or more attached cover letters (one per company/organization which you are targetting). Follow these steps:

1. Choose one or more projects to which you wish to apply.
2. Submit an e-mail to smpindustry@uq.edu.au with the title:
   INTERNSHIP APPLICATION: --Name--, --Student Id--
3. In the e-mail specify the specific company/organization to which you are applying (as per the projects below).
4. Attach your CV (pdf file) to the e-mail. Name the file YOURFULLNAME_cv.pdf. For example: JessicaWu_cv.pdf.
5. Attach your latest UQ grade transcript to the e-mail. Name the file YOURFULLNAME_grades.pdf. For example: JessicaWu_grades.pdf.
6. For each company/organization to which you are applying, attach a cover letter (pdf file). Name the file YOURFULLNAME_COMPANYNAME_cover.pdf. For example: JessicaWu_FlightCentre.pdf. Your cover letter should not exceed one page in length and may ideally be 2-3 paragraphs long. Your cover letter should address the company/organization to which you are applying and may reflect upon your potential suitability. Keep in mind that you are not expected to be a professional at this point - it is an internship.
7. Applications close on September 30, 2017.

All applications are then sent out to the respective companies/organizations. The companies then shortlist applicants for phone/upfront interviews during October 2017 with a closing date of October 20, 2017. Successful applicants may then accept internship offers by November 3, 2017. At that point, sign-on to SCIE3255 is also possible, where applicable.

---

Available Projects

Notes: Projects are located in the Brisbane area unless stated otherwise. Some companies/organizations feature several projects - in such cases applications are for the whole company/organization (not for individual projects).

Aginic is a Brisbane based Data Analytics startup. We were founded in July 2014 and have grown to over 40 employees across the east coast of Australia. We are a young and energetic team that work with large organisations to solve problems using data. As part of your internship you will join a team on a real data analytics project and work on technical development such as data extraction and modelling as well as visualisation. You will be surrounded by a supportive team that will help you learn and guide you through the process.

Additionally we are looking for similar interns for our AginicX app development area of the business. A project in this area would involve contributing to an Agile development team to build a real-world mobile or web app for business or public consumption.

Desired skills: Maths, Engineering.

To assist Australian University Sport in streamlining their sporting results processes within the domestic and international programs. Involves statistical analysis and collation of information into usable formats.

What we want to achieve: streamline our results management processes including developing a searchable results platform that is easy to update and report from. Project will include collating current results within the organisational files and streamlining how these results are kept and maintained with view to ensuring they are accessible to an online searchable results platform. Project will also enable the person to contribute to improving the organisations current championship results spreadsheet using their knowledge of excel mathematical formulas.

The location of the internship can be remote or the person can work from the AUS office located in Milton or on the Gold Coast.

Desired skills: Knowledge of sport and interest in sport statistics/results would be an advantage. Excellent knowledge of excel formulas would be welcomed!

1. Monitoring of Predictive Models: Develop an automated tool to alert the modelling team when a model's predictive power is deteriorating and requires a refresh or rebuild.
2. Risk Concentration Monitoring Program: Develop a monitoring tool to provide alerts for geographic concentrations of risk, specifically for natural perils.
3. Develop a "Call Model": Develop a model to predict which customers are likely to call the insurer as a policy comes up for renewal. Understand how this behaviour can be influenced and build this insight into the way renewal prices are set.
4. Develop an "Excess" Model: Develop a model to predict the Excess levels selected by a policyholder and impact of the selection on the underlying risk characteristics of the customer.
5. Predict risk using vehicle crash data. Use publicly available crash statistics to predict claims experience.
6. Develop a model to predict quote behaviour Use big data techniques to analyse and understand the quoting behaviour of customers and potential link to underlying risk characteristics of the customer.

Desired skills: An understanding of statistical / mathematical model building techniques, programming algorithms in SAS, R or other relevant tools, including development of appropriate metrics for on-going monitoring.

We are rapidly increasing our storage of events for everything that occurs in Autopilot, from contact updates, automation events, email (clicks, opens, delivered, etc), page clicks w/ UTMs, headsup events, form submissions, custom events, etc. etc. We currently store 10s of millions of events per day and expect this to rapidly increase as time goes on. Additionally, we are opening up our v2 API which will allow clients a much greater variety and volume of data they can store in Autopilot. This includes relational, time series and document-based data. For this purpose we are developing a data layer which will allow us to control where and how we store this data and change that quickly and cheaply. There are several things we want to accomplish on a statistical/ML front with this data:

• Industry insights, i.e. "People in your industry who do X get % response"
• General insights into the data - what trends do we see, what interesting patterns are there?
• Can we give people ideas of how to build journeys based on the data we are recording?

Desired skills: Data Analysis, statistics, machine learning and programming.

We are a Commercial Mathematics company based in Australia and working with customers internationally providing quantitative solutions to complex business problems. We use AI techniques from Operations Research, Data Analytics and Machine Learning to build mathematical optimisation engines for customers in a wide variety of industries.

We have vast sums of customer data that capture their operations and use of our tools that we think can be utilised to provide valuable insights into the way they manage their business operations and their usage of our tools. We would love to bring in someone with deep statistical and mathematical modelling skills to analyse this data and draw real-world conclusions to help us better support our customers in a consultation manner and deliver improved tools.

We are open in terms of approach, techniques, tools, etc. The point of contact will help with access to data, systems, subject matter experts and so on.

Desired skills: Statistical analysis, mathematical modelling, data science (data munging, transforming and cleaning), a passion for diving into the data and drawing real world conclusions.

We are a wholly owned subsidiary of CoreLogic U.S. (NYSE: CLGX), the largest data and analytics company in the world. CoreLogic provides property information, analytics and services across Australia and New Zealand and is currently developing and growing partnerships throughout Asia. We are the leading provider of property data and analytics services for consumers, investors, real estate, mortgage, finance, banking, insurance, developers, wealth management & government.

We have a veritable trove of customer activity data that captures the who, what and when of over fifty APIs that provide data, statistics and insights for our applications, customers and third-party integrators. We would love one or two people with deep statistical and modelling skills to analyse this data (and its related real-estate reference data) and provide insights into how our various types of customers use the APIs. This will help us to build better and more intuitive APIs.

1. We have scope to work with two people, which would have the advantage of allowing both people to bounce ideas off each other; 2. We are open in terms of approach, techniques, tools, etc. The point of contact will help with access to data, systems, subject matter experts and so on.

Desired skills: statistical analysis, mathematical modelling, data science (data munging, transforming and cleaning), a passion for diving into the data

Project 1: Conquering weed problems on farms using drones and machine learning.
Invasive weeds are one of the biggest threats to Australia's natural environment and agricultural industry. The aim of this project is to develop a method for mapping invasive weeds by combining drone imaging technology with the latest machine learning techniques. In this exciting project, you will assist with multi-rotor or/and fixed-wing drone data acquisition, learn how to process drone imagery and train machine learning algorithms to detect and classify different species of weeds to generate an accurate weed distribution map of farm properties.

We have the scope to work with two students. Depending on their interest, it will be possible to focus on either drone imagery pre-processing or developing approaches for automated weed detection using machine learning. Within the project, students may be required to travel to field sites near Cairns.

Project 2: Applying Recurrent Neural Network and Statistical Techniques on Water Quality Analysis.
Good water quality is the reason why the Great Barrier Reef is one of the most beautiful, diverse and complex ecosystems in the world. However, the quality of water entering the reef has deteriorated over the past 100 years and continues to have a detrimental effect on the marine ecosystem. As part of the Great Barrier Reef Protection Project, this project aims to use advanced machine learning and statistical techniques to predict water quality (nitrogen discharges). The student involved in this project (within the box in Figure 1) is planned to collect different data streams from SensorCloud system, implement various statistical analysis on the time series data (such as nitrogen, temperature, rainfall), build and run a RNN model based on Tensorflow (or other ML platforms) to predict the water quality. From this project you can gain experiences of a complete ML analysis workflow and contribute to the Reef water quality protection by bringing professional understanding and mathematic skills to the time series data analysis work.

Desired skills: Programming skills (prefer Python); knowledge about time series data processing; no specific ML or engineering knowledge required, but it is good to have solid knowledge about probability and linear algebra.

This project will evaluate the performance of a low cost water quality system (Nitrate and chloride) providing information on sensitivity, selectivity, drift and other factors of performance. The output from this project will be a report on the performance of the low cost water quality system, including a robust statistical analysis. The student, working with technical support and research supervision will assist installing the system in a convenient location. The system will be connected to a data retrieval system and monitored for performance. The student will develop a experimental design for the assessment selecting appropriate statistical methods for the analysis. Independent water quality samples will be taken at times developed with consideration of dynamics and variability. These samples will benchmark the performance. The assessment will be recorded in an output report.

Desired skills: An interest in environmental problems; Experimental design; Statistical analysis; Report writing.

Evolving Machine Intelligence (EMI) is a tech startup developing new forms of evolutionary Artificial Intelligence (NOT neural networks/deep learning; we do more interesting work than that). Although this largely involves applied mathematics and computer science, some questions with a pure mathematics flavour arise from time to time, which we would like to investigate further. Dominant topic at the moment: consider an arbitrary (assume, well-behaved) dynamical system defined by a set of nonlinear ordinary differential equations, the parameters of which are within a specified compact hypercube in parameter space. Investigate the attractor of the system as evidenced by trajectory behaviour in phase space. How is the behaviour of the attractor to be quantified? Given any arbitrary countable set of candidate parameter vectors selected from the hypercube, how do we describe the influence of the attractor on the behaviour of the dynamical system in phase space space for these candidate vectors? Now, working backwards: if we know a specified arbitrary segment of phase space trajectory of the system, how do we remove (or at least, quantify) the influence of this attractor when performing a random heuristic search to find candidate parameter vectors for the dynamics? How is this task altered if the attractor is Strange? (Remembering that it is possible to have a Strange attractor for a non-chaotic system.) What is the most efficient way to express this information mathematically?

Desired skills: Ordinary Differential Equations; some knowledge of complex systems plus a more general background in pure mathematics and basic programming (preferably Matlab); some applied mathematics is also good.

Flight Centre Travel Group is offering up to two positions for strong UQ SMP students to work within the Data Science Unit, a component of the Customer Intelligence and Analytics team. As Australia's leading travel solutions company, we're taking a data-driven approach to decision making, applying advanced predictive modelling and analytics to large datasets to obtain valuable business insights. Through this internship you will gain experience with one or several of: Enterprise-level database systems, getting and cleaning data, exploratory data analysis and visualisation, creating predictive models through machine learning and other techniques, documentation and business processes.

Desired skills: Solid rudimentary statistics understanding. Programming experience viewed favourably, particularly experience in SQL, R, and dealing with datasets. Nice to haves: C#, DBA experience, experience creating dashboards.

Aerial lidar surveys generate large amounts of unstructured three dimensional point cloud data. Automatically extracting a correct model of the terrain height from such a point cloud is a challenging algorithmic problem. At Fugro Roames, we have a carefully hand tuned system for doing this, but it still occasionally makes mistakes, for example, considering very large buildings to be part of the terrain. In this project, you would use supervised machine learning techniques to reduce the error rate on diverse and interesting data sets - for example, a full scan of Brisbane city.

Desired skills: You really need some existing programming skills to make this a successful project. Ideally you would have some C++ skills, along with experience in a technical computing environment such as julia, matlab or scipy. You also need a strong quantitative background. Nothing specific is required here, but the task requires a lot of three dimensional geometric reasoning, and the ability to express this as numerical code.

BOLT is a multi-modal supply chain optimisation tool developed by Polymathian. BOLT optimises the operations of supply chains: spanning from strategic decision making to operational execution. Polymathian is looking for students with software development experience to work with BOLT on data analysis, visualisation and interpretation.

Desired skills: Software development experience (Python is ideal, knowledge of object oriented programming is a must) Will be viewed favourably: Statistics; Operations Research.

With headquarters in Brisbane and listed on the ASX (RUL), RPMGlobal is a global leader in the provision and development of mining software solutions, advisory services and professional development. Our global team has over 45 years of experience in the mining industry and are the largest group of independent technical experts in the world.

1. Participate in the ongoing development of our XERAS Enterprise software. Contribute to the development of "Construct mode" for XERAS Enterprise in an agile software development environment. XERAS Enterprise provides an Enterprise Financial Modelling system that keeps operations on course and in control of costs.
2. Participate in the ongoing development of our XECUTE software. Contribute to the development of the next release of RPM's ultrashort term scheduling product XECUTE in an agile software development environment. XECUTE combines design, reserving and scheduling in a single, dynamic, 3D software tool.
3. Participate in the ongoing development of our mine scheduling software products. Contribute to the development of the next release of RPM's mine scheduling products in an agile software development environment. XPAC Solutions include an optimised scheduling core with advanced scheduling logic that utilises multiple solvers including Gurobi and Lindo.
4. Participate in the next phase of augmented and virtual reality innovation at RPM. Contribute as a member of an internal project team which has the objective to integrate 3D visualisation and animation in RPM products with Apple ARKit and Google ARCore. The solution is anticipated to be achieved via integration and extension of RPM's Visualisation Library with Unity.
   
   Desired skills: Interest in financial data modelling/analysis and software engineering using C# (strong C++ or Java may be a suitable substitute), interest in a technical software engineering role, strong interest in AR and VR, knowledge of 3D programming, some experience using Unity, experience developing for an iOS or Android device.

Scientific Games is a leading innovator in the regulated gaming and global lottery industries. We specialise in the development, production and distribution of fun and exciting products such as Instant Scratch products and systems, Electronic Table Games, Gaming Machines and advanced Casino systems. Beginning with the breakthrough technology that launched the world's very first secure instant lottery game in 1974, Scientific Games has continued to be a driving force behind the success of more than 300 customers on six continents over the last 40 years across 3 divisions – Gaming, Lotteries and Interactive.

In the long history of the gaming industry a broad variety of poker machine game designs have been created but a limited number of mathematical tools and methods have been developed to examine them in detail. The sort of problems that arise in this area are simple to describe but have non-trivial solutions. We would love to introduce someone with an interest in games and a firm understanding of combinatorics and probability to our field. This will help us identify underlying structures of these games, build better mathematic models, and make games which are more fun.

Desired skills: probability, combinatorics, curiosity and a passion for games (of any kind). You will be joining a small team and will take part in regular discussion and brainstorming sessions on game and mathematical topics.

Additional Information: Scientific Games are prepared to cover 2 return trips from Brisbane to Sydney, but not transfer costs from the airport such as trains, taxis, uber, bus etc. Successful applicant will be responsible for finding their own accommodation during the internship.

SilverRail is dedicated to making Rail travel seamless by providing state of the art software to consumers to help them solve their travel and ticketing problems. The successful intern will have the opportunity to choose a project in conjunction with their supervisor to match their unique skills and experiences. Examples of projects include creating software based around cutting edge networking and optimisation research, writing tools to provide sophisticated analysis and visualisations of complex industry data, machine learning techniques to enhance and improve the selection of responses given to customers, travel apps and novel ways to apply graph theory to real life problems

Desired skills: Mathematics and programming ability.

The McKinnon Group is a company that owns a number of different companies. This specific project is for our childcare software group although we also own a number of childcare centres. The Australian Childcare industry is an 11 Billion dollar industry annually with over 10,000 centres across Australia. Wages in those centres usually run at between 50-60% of revenue (~5 - 6 billion dollars) and it employs 172,000 people.

The advent of electronic sign in / out sheets into the childcare industry over the last 6 months has created an opportunity for our software group to be smarter in the allocation of rosters. There is currently no childcare specific rostering package but adaptions of generic packages. We will be able to accurately know to the minute when parents are turning up with children. It is estimated in our centres that our staff are turning up 30 minutes to 1 hour earlier than required which adds 6-12% additional cost to the overall childcare industry or ~300-600 million in wages per annum which is passed onto parents.

This project is to create a predictive model that will be used in commercial software to proactively allocate future rosters to childcare workers.

Desired skills: Mathematical and programming ability.

Quantitative analysis of real-world market data, latency-critical system performance or other data analysis. Tibra is high performing organisation in an extremely fast moving industry, so a precise description of intern tasks is difficult.

Tibra Capital is a quantitative research and investment group that utilises its shareholder's capital across a broad range of investment strategies. Our mission is "to facilitate liquidity, price fairness and efficient allocation of capital in financial markets, leading to growth in our investors' wealth" and we value diligence, innovation and integrity.

Our investment philosophy focuses on a scientific approach using a range of systematic and diversified trading strategies across all asset classes, and superior risk management techniques to manage the exposure of our investments. We use a mix of statistical and fundamental analysis to gain an edge for our investment strategies.

Desired skills: We seek out candidates who can demonstrate passion and talent for developing algorithmic solutions to quantitative problems. Some programming experience is essential. Applied Mathematics, Statistics, Machine Learning and Numerical Modelling are all relevant skills.

---

Doing it For Credit

If you are accepted to an internship, you may opt to take the internship for credit by signing on to SCIE3255 for Summer Semester 2017. To do so, contact Yoni Nazarathy to request to signup for SCIE3255 (only after being offered an internship). At that point, the internship project will be further reviewed by academic staff to ensure that it aligns with the learning outcomes of SCIE3255 and if approved you may sign on to the course.

As a credit course, the internship course is a pass/fail course. Assesment involves four items.

• Weekly reports - A short weekly report (few lines) indicating the activities carried out..
• Intermediate report - An intermediate report occuring prior to the "withdraw without academic penalty cutoff date".
• Final report and presentation.

Further details about assesment is here.

---

Frequently Asked Questions

Q: Can I participate as an intern if I am a Bachelor/Honours/Masters/PhD student?
A: The program is open to Bachelor, Honours and Masters students but not to PhD students this year.

Q: I am about to graduate on November/December. Can I still participate?
A: In general no. An exception is if you are a Bachelor student starting another program of study in the school (e.g. Honours or Masters) in the immeidate following year.

Q: Can I do the internship for credit if I am a ______?

• Bachelor Student.
       A: Yes, provided that you are not graduating now.
• Bachelor Student about to start Honours.
       A: No.
• Bachelor of Advanced Science Student.
       A: Yes, but this course cannot count towards your Honours component.
• Honours Student.
       A: No.
• Master Student.
       A: No.

Q: How long is the internship?
A: Internships are 6-8 weeks in length. An insurance restirction implies that at most four weeks can take place per calendar year (e.g. 4 weeks in 2017 and 4 weeks in 2018). Internships take place during the months of November, December, January and February with the specific schedule determined by the industry partner.

Q: Is it possible to undertake aditional academic study during the summer semester?
A: It is not advisable, but it is technically possible. For example - if the dates allow for it, you wish to attend the AMSI Summer School.

Q: Is the course SCIE3255 (this internship project) compatible with other internship courses such as research internships at UQ?
A: Yes - your degree may include this internship together with other types - but for the summer duration during which you do the internship, it is not possible to do both.

Q: How much do I get paid?
A: You will be paid an Honoriarium of $360 per week for each week of internship activity, but note that the payment that you receive will not be provided on a weekly or fortnightly basis: you will receive a block payment at the end of the 2017 period of the internship, and another at the end of the 2018 period of the internship.

Q: If I am doing it not for credit, in addition to working with the industry partner, what does UQ require from me?
A: You are required to partcipate in an event taking place during the first weeks of Semester 1. You may be asked to prepare a talk for this event.

Q: If I am doing it for credit, what does UQ require from me?
A: See the for credit section above.

---

Further Resources

• UQ Employability Page - contains many resources useful for career planning, job searching, cv preperation, the interview process and the like.
  
• UQ Idea Hub - Have your own startup idea? Go here.
  
• UQ Employability award.
  
• Employ101x MOOC.
  
• Graduate Employability 2.0.
  
• Something a bit different: Didn't get an internship but want to finish the summer a whole lot smarter and more employable? Pick up some data science. Perhaps start here and see where watching these videos takes you.