London, United Kingdom

Prashant Verma

Senior Data Scientist

Data science for Coca-Cola, Unilever, NBCUniversal, BT and EE. I take ideas from a rough first sketch all the way to production.

About

Five years of taking ideas from a rough sketch to something running in production at global scale.

I've worked across media, telecoms, FMCG, healthcare, education and finance, usually as the person pulling together data science, engineering, marketing and ops teams and keeping them pointed in the same direction. Right now I'm a Senior Data Scientist and Project Lead at WPP Media, working with clients like Coca-Cola, Unilever, NBCUniversal and BT.

I enjoy solving business problems with a bit of creative thinking. Sometimes that means building a model, sometimes it's building the right tool, and sometimes it's just having the right conversation with the right person. I get as much satisfaction from the technical build as I do from reading a room and working out what a client genuinely needs.

Increasingly the work is less about writing the models myself and more about setting direction, mentoring the people who do, and translating all of it into decisions non-technical leaders can act on.

Experience
Since 2024 · London

Senior Data Scientist, Project Lead

WPP Media
Coca-Cola · Unilever · NBCUniversal · BT · EE · Welcome Break

I lead projects across roughly 25 cross-functional stakeholders spanning data engineering, data science, marketing and MLOps, taking solutions from ideation to production.

  • Architected CLV prediction and audience optimisation models powering Google Ads bidding and cross-sell campaigns, delivering a 20% POAS uplift with lower acquisition costs across broadband and mobile.
  • Built and automated a media mix response model spanning 23,500 brand-markets across 117 countries, used by Unilever markets to plan annual media budgets.
  • Drove geo-spatial audience strategy and measurement (DiD, Causal Impact) for Coca-Cola's FIFA World Cup campaign across 20 markets, and contributed KNN statistical matching to WPP's Open Intelligence data fusion platform.
  • Engineered a multi-agent GenAI system with BigQuery function calling, a GenAI layer turning MMM outputs into monthly effectiveness reports for EE, and finance report automation at NBCUniversal.
  • Built ETL pipelines processing 30M BT and EE customers plus real-time value-based bidding with sGTM, Firestore and Vertex AI. Designed EE Marketplace's taxonomy and recommendation engine, saving 1 FTE.
BigQuery · Terraform · Airflow · Vertex AI · Firestore · GenAI / LLM · MLOps · CI/CD
2023 to 2024

Data Scientist

Twinkl Ltd

Worked alongside merchandising, marketing and finance on the commercial side of a global education business.

  • Customer segmentation, cohort analysis, lifetime value and A/B testing that delivered £348K in revenue growth.
  • A reproducible Prophet forecasting model predicting subscriptions and revenue at roughly 85% accuracy, feeding quarterly budget allocation.
  • SpaCy language models detecting approximately 28,000 spelling errors across 10 languages.
  • A customer journey strategy built from behavioural patterns that lifted retention 10% by preventing cancellations and driving upgrades.
BigQuery (GA4) · MySQL · Tableau · Looker Studio · Prophet · SpaCy
2022

Data Scientist

Moorfields Eye Hospital

Clinical NLP research, working directly with ophthalmologists to make free-text patient notes machine-readable.

  • Compared RNN, LSTM and BERT for entity recognition in a hospital setting, a 30% accuracy improvement over traditional methods.
  • Ran named entity recognition and relation classification across 5 million patient records to surface ocular health insights.
  • Partnered with clinicians to label 6,000 clinical letters, and shipped the tooling as React and Django web applications.
BERT · LSTM · NLP / NER · Label-Studio · React · Django
2021 · Gurugram

Data Scientist

Sutra Tech Labs

My first data role, and the one where I learned to lead a team rather than just contribute to one.

  • SQL querying and wrangling to identify popular keywords and integrate hashtags into news feed content, lifting content relevancy and user satisfaction by 30%.
  • Managed marketing and design teams on data-driven outreach, a 25% increase in user engagement.
  • Directed a 6-member team building an AI-powered virtual police station for the Government Police of India, delivered in six months and benefiting 82.3 million people.
SQL · Python · Django · Machine Learning · NLP
Selected work

Machine learning, NLP and full-stack projects, including work with the Madhya Pradesh Police and India's Ministry of AYUSH, live on github.com/Prashantvermapv

Beyond the day job
Co-Chair and Community Manager, WPP Unite
Leading employee resource group initiatives and managing a team to deliver inclusive culture programmes across WPP Media.
Senior Postgraduate Teaching Assistant, UCL
Supervised, led practical sessions and graded 400+ business undergraduates across database systems, web technologies, Java and Python.
Data science volunteer, DataKind and SEAPIA
Monthly volunteering with 3+ social change organisations, building their data capability and confidence in data-driven decisions.
Research assistant, UCL and UoW
Developed two machine learning models for healthcare in university research labs.
Hackathon wins
1st prize (₹100K) for VCop, beating 132 teams and 792 participants in 2021. 1st prize (₹50K) for VAIDYA, in partnership with the Ministry of AYUSH.
Skills and tools
Python
Pandas, Polars, NumPy, scikit-learn, TensorFlow, PyTorch, Keras, PySpark
NLP and Generative AI
BERT and Transformers, SpaCy, NLTK, HuggingFace, LLMs, Gemini, Claude, multi-agent systems
Cloud and data engineering
Google Cloud, BigQuery, Terraform, Airflow, Vertex AI, Firestore, sGTM, CI/CD, AWS, Azure Databricks
Databases
MySQL, BigQuery, PostgreSQL, Neo4j, MongoDB
Visualisation and BI
Tableau, Looker Studio, Power BI, Matplotlib, Seaborn, Plotly
Web
React, Django, Flask, JavaScript, HTML/CSS
Ways of working
Stakeholder management, project leadership, mentoring, experimentation and A/B testing, causal inference, agile delivery
Education and certifications
MSc Data Science and Health
2022
University College London
Distinction
B.Tech Computer Science
2021
Maharaja Agrasen Institute of Technology
9.3 / 10
Professional certifications
  • Google Cloud Certified: Professional Machine Learning Engineer
  • Google Cloud Certified: Professional Data Engineer
  • Google Marketing: Analytics, Ads Search, Social Media Marketing, Media Planning, Activation, AdOps and Media Investment
Contact

Always happy to talk data strategy, GenAI in practice, or how marketing can be a bit cleverer.

Emailprashant2798.pv@gmail.com LinkedIn/in/prashantvermaa GitHubPrashantvermapv RésuméDownload ↗
↑