Data Scientist

The purpose of this role is to lead and manage customer analysis deep dives, execute the strategies outlined by senior leadership, and manage the day-to-day communication with clients.

Job Description:

Responsibilities
1. Creating and developing marketing measurement solutions (e.g. media mix modelling, customer lifetime modelling, attribution and incrementality testing tools) using machine learning and probabilistic programming techniques.
2. Developing code for our solutions using Python libraries such as PyMC, PyTensor, NumPy, SciPy and StatsModels.
3. Maintaining Github repositories and cloud computing resources for effective and efficient version control, development, testing and production.
4. Researching potential new methodologies using academic and non-academic materials.
5. Productionising proof-of-concept solutions and assisting in rolling these out to our clients.
6. Producing and updating documentation and training materials.
7. Supporting team members in the UK and internationally with adopting and implementing solutions on client accounts, as well as receiving and actioning feedback and feature requests.
8. Working with data engineering and UX teams to jointly develop end-to-end solutions.
9. Working with business development and client account teams to help shape the commercial model and “go-to-market” proposition for the solutions we develop.
Must have:
1. Coding using Python, including use of common packages such as Jupyter, NumPy, SciPy, Pandas, Matplotlib and Seaborn.
2. Solid mathematical knowledge, across topics such as algebra, calculus, trigonometry, and graph theory.
3. Machine learning theory and application, such as regression, classification, and cross-validation.
4. Parametric statistics, probability theory and principles of Bayesian inference.
5. Media mix modelling concepts, such as adstock and diminishing returns.
6. Basic time series modelling techniques, such as ARIMA.
7. Basic optimization concepts, such as Lagrangian multipliers, constrained vs. unconstrained optimization, local vs. global minima.
Good to have:
1. Probabilistic programming Python libraries such as PyMC or Tensorflow Probability.
2. Collaborative version control software such as Github.
3. Cloud computing software such as Google Cloud Platform or Databricks.
4. Probabilistic/Bayesian modelling approaches, such as hierarchical regression, counterfactual inference, mixture modelling, factor analysis and Gaussian processes.
5. Advanced time series modelling techniques, such as Fourier regression, vector autoregressive (VAR) models and stochastic differential equations (SDEs).
6. Markov chain Monte Carlo (MCMC) methods and their application in probabilistic programming.
7. Optimization methods, including single- vs. multi-objective, deterministic vs. stochastic, local vs. global and gradient-based vs. gradient-free optimization algorithms.

Location:

DGS India - Bengaluru - Manyata N1 Block

Brand:

Merkle

Time Type:

Full time

Contract Type:

Permanent