Data Scientist – FEU Development & QA

Key Skills

Aartificial intelligence, Agile methodologies, Algorithms, Azure, Databricks, Deep Learning, KPIs, Machine learning techniques, MLOps tools, Project management, Python Programming, Spark Core, Statistical modeling

Job Description

The Data Scientist is an exciting role in the Data Science Centre of Excellence (CoE) team within D&A. This team builds state of the art machine learning algorithms, maximising the impact of D&A’s solutions in driving enterprise performance. Typical initiatives include building a state of art enterprise scale demand forecasting engine, optimizing trade promotion investments, accurately forecasting customer demand, using NLP to glean insight on consumer trends from search data, and making individual assortment recommendations for each of the millions of stores that sell Unilever products. The role will work closely with a D&A functional team such as Customer Development or Supply Chain along with business stakeholders, addressing both large transformation programs and near-term business problems.

Main Purpose of the Job:

(A concise statement setting out the main purpose and objectives of the job)

Forecast engine utility (FEU) is the Unilever’s inhouse AI/ML based demand & promotion forecasting solution. Data Scientist role will be responsible for regional development of this technology and own the forecast KPIs for the respective region (group of countries). Key Responsibility of this role will be to a data science design authority during the global forecasting solution rollout for the region.

Key Responsibility:

• Develop and Govern data science build process and model Change landscape for enterprise level one-forest solution together with business team such as digital supply chain, integrated customer development team and Build partners
• Technical due diligence of market DS work. Advanced knowledge on building and implanting scalable forecasting solution
• Perform FEU forecast model diagnosis during development phase
• Track model features and periodically evaluate model performance KPIs
• Identify key gaps early on in development on data, feature or model which may lead to poor performance in parallel run
• Push global DS model guideline/governance during project phase
• Identify and proposed continuous improvement plan from FEU re-usable module (FEU LEGO)
• Training and supervising build team on global forecasting library (LEU-LEGO) and strengthening technical gaps in global library by customizing the same as per region specific data and model requirements.
• Write efficient and well-organized ML based software to ship products in an iterative, continual-release environment.
• Engaging in strategic discussions to identify problems while measuring and improving the quality of workflows. Identify and implement opportunities to improve tooling and implementation of scalable solutions
• Working in an agile environment with globally distributed teams.

To do this job successfully, you will need strong skills in Timeseries models, Machine Learning, and strong hold of object-oriented programming with large data. Experience in enterprise level ML product development is required.

Key accountabilities:

(Describe the decisions this role is accountable for and the responsibilities/deliverables expected)

• Development & Deployment of state of art ML/AI models for timeseries demand forecasting
• Evaluation of FEU market models as per global data science design
• Provide new solutions to enhance the FEU demand forecast for new products, baseline, promotion lift and total forecast
• Lead pilots and experiments to improve forecasting performance and scope
• Implementation of scalable data science solution with market explainability features
• Manage forecast run performance to achieve forecast accuracy, bias and no touch adoption targets
• Responsible for technical strengthening of market data science models
• Interact with relevant teams to identify business challenges where data science can help
• Apply comprehensive data science knowledge to propose optimal techniques for key business challenges
• Create detailed data science proposals and project plans, flagging any limitations of proposed solution
• Design and prototype experimental solutions, particularly machine learning models
• Facilitate industrialization and ongoing operation of solutions through well organised code, clear documentation and collaboration with ML Ops resources
• Represent Data Science in cross-functional governance of projects, engaging with stakeholders up to Director level
• Highlight recent developments in data science capability which could solve additional challenges
• Lead a team of up to 1-2 data scientists / interns (WL2A), providing career mentorship and line management
• Provide technical guidance to data scientists across D&A, particularly on the projects you lead
• Support the growth of D&A’s data science capability by contributing to activities such as tool and vendor selection, best practice definition, recruitment, and creation of training materials
• Build the reputation of D&A’s data science capability within Unilever and externally, through activities such as community engagement (e.g. Yammer), publications or blogs
• Provide ad-hoc & immediate support to the business when needed (for example Covid-19 crisis support)

Depending on the specific project, the Senior Data Scientist can expect 80-90% of their work to be hands-on prototyping solutions, with the remainder spent planning and designing, overseeing and reviewing work of project staff, interfacing with stakeholders and managing team members.

Experience and qualifications required:

(Detail essential and desirable experience)

Standards of Leadership Required in This Role

• Personal Mastery (Data-science and advanced analytics)
• Agility
• Business acumen
• Passion for High Performance

Key Skills Required

Professional Skills

• Machine learning Expert
• Statistical modelling Expert
• Forecasting Expert
• Explainable ML/AI Expert
• Python/Pyspark coding Expert
• Data science platform tools e.g. MS Azure, Databricks Fully Operational
• Deep learning (temporal & spatial learning ) Fully Operational
• Collaborative development using Git repos Fully Operational
• Automated Machine Learning platforms Foundational knowledge

While a broad data science technical background is required, the role will benefit from deeper skills (for example graduate studies or prior work experience) in one of the following areas, optimization, simulation, forecasting, natural language processing, computer vision or geospatial analysis.