Brandwatch is the global leader in social and digital consumer intelligence, working with the biggest brands and most admired agencies worldwide – 39 of the Fortune 100 are clients.
Recently named a leader in The Forrester Wave™, we build smart software solutions that help capture, understand and share insights from millions of conversations across the web and beyond. Home to over 500 people worldwide with offices in 10 locations, Brandwatch is a place that motivates and supports its staff to find answers to new problems in an ever changing social and digital world.
Title: Data Science Intern
Location: Remote UK
Terms: Full time, temporary, 3 month contract / 12 weeks (with potential to extend)
This internship is an opportunity to join the Data Science team at Brandwatch to assist in the delivery of various machine learning and predictive analytics projects. The intern will be responsible for the quantitative evaluation of our in-development classifiers and models. The role will involve collecting and analyzing a large amount of gold-standard evaluation data, with the goal of assessing the performance of our existing models and to inform the direction of future enhancements.
Background on the role
Brandwatch builds a suite of analytics software products that allow our customers to measure and predict consumer behaviour online. We ingest enormous quantities of unstructured consumer voice data; such as publicly available social media data, blogs, forums etc.... Our customers then search and analyze these data, using the tools we have built, to achieve various marketing and market research outcomes; for example: “What do consumers like about our new product?”
The role of the Data Science team is to design new products and features that support our users by automating and streamlining their work. We draw on our experience in machine learning, natural language processing, and statistical analysis to make sense of all our data. Example projects include detecting emerging crises, identifying the key topics of conversation, and measuring opinion. A critical part of all this work is developing robust evaluations of our systems, to discover how well we are currently performing and to help prioritise future enhancements.
This internship is an opportunity to join the Data Science team, to support us in our evaluation efforts. The successful candidate will work closely with senior team members to design and implement robust quantitative studies against one of our models.
Responsibilities and Duties
The internship will be a single specific project: The end-to-end delivery of an evaluation against one of our machine learning classifiers. Such a project has many stages, and so we anticipate that the internship will cover a range of tasks and duties. These may include:
- Technical writing: Authoring detailed project plans, to get as much up-front clarity about the project as possible. Communicating the results of the work with stakeholders. Writing guidelines for human annotators of the data.
- Data exploration and processing: Collecting a very large corpus of social media data from our platform. Filtering the corpus such that we achieve a representative sample for the given task.
- Manage the annotation workforce on our crowdsourced annotation platform, to ensure guidelines are adhered to and quality standards are met.
- Data analysis and modelling: Collate and aggregate the annotated data. Apply statistical weighting to the evaluation data to produce various quality metrics of overall system performance. Model the evaluation data to produce generalised insights of how our systems perform, to set expectations and to inform future enhancements.
Mentoring and guidance will be available for every stage, but it is our expectation that the intern will complete the majority of the work themselves.
At a minimum a successful candidate will have some prior experience in Python programming for data exploration and analysis. Beyond that, other role specific skills can be learned on the job, though prior experience in any of the following areas will be beneficial:
- Standard Python frameworks for Data Science: SciKit-Learn, Pandas, NumPy
- Distributed/cloud compute: AWS, EC2, S3, PySpark, Dask
- Human annotation methodologies and metrics: Inter annotator agreement, Cohen's kappa.
- Machine learning evaluation techniques and metrics: Confusion matrices, Precision, Recall, F1, ROC-curves and AUC.
- Statistics sufficient for data analysis, re-weighting, and predictive modelling.
We offer: Flexible & remote working, a competitive benefits package, an extensive people development program, including LinkedIn Learning for every employee and in-house courses such as our Brandwatch Leadership Development program, and community-focused groups, such as our Diversity & Inclusion and Green committees.
Working at Brandwatch: Our values are about being authentic, bold, and creative. And we believe these values are best embodied by enabling people to do their best work in an environment that works for them. Whether fully remote, within one of our global offices, or a combination of the two, we have the tools and resources to make working to these values both possible and enjoyable.
To all recruitment agencies: Brandwatch does not accept agency resumes. Please do not forward resumes to our jobs alias or Brandwatch employees. Brandwatch is not responsible for any fees related to unsolicited resumes.