Full-time · Remote (India)

AI Marketing ROI Forecasting Scientist

Cehpoint is hiring for this role: full-time, Remote (India), ₹12,00,000 – ₹20,00,000 / year. Build ML-driven forecasting models that measure and predict marketing spend ROI across enterprise campaigns.

About the role

As an AI Marketing ROI Forecasting Scientist at Cehpoint, you will own the end-to-end development of predictive and causal models that quantify how every marketing dollar translates into measurable business outcomes. Your work will directly shape multimillion-dollar budget allocations across our enterprise clients, bridging rigorous statistical science with actionable growth insights.

What you will do

  • Design and implement marketing mix models (MMM), multi-touch attribution (MTA), and uplift/causal inference models to forecast campaign ROI and incrementality.
  • Build and maintain scalable feature pipelines ingesting ad spend, clickstream, CRM, web analytics, and macroeconomic signals from cloud data warehouses.
  • Conduct rigorous experiment design, power analysis, and causal identification strategies; validate model predictions against holdout markets and randomized campaigns.
  • Develop Bayesian hierarchical and probabilistic ML models to quantify uncertainty, detect diminishing returns, and segment ROI by channel, geography, and audience cohort.
  • Translate model outputs into executive-ready dashboards and narrative reports, partnering with marketing leadership to convert forecasts into budget-recommendation playbooks.
  • Stay current with frontier research in marketing science, LLM-augmented attribution, and privacy-safe measurement frameworks; prototype and productionize promising approaches.

What you should know

  • Causal inference foundations — potential outcomes, propensity scoring, instrumental variables, difference-in-differences, and their marketing-science applications.
  • Bayesian statistics & probabilistic programming — hierarchical models, MCMC/VI inference, posterior predictive checks, and uncertainty quantification for decision-making.
  • Marketing science fundamentals — media-mix modeling, brand-lift experimentation, funnel attribution, and the interplay between short-term performance and long-term brand equity.

What you should be able to do

  • Python — scikit-learn, XGBoost/LightGBM, Prophet, PyMC/Stan, statsmodels
  • SQL & cloud data platforms — BigQuery or Snowflake, dbt for transformation layers
  • Experimentation tooling — R or Python for A/B-test analysis, power & sample-size calculations
  • Visualization & communication — Python (Matplotlib, Seaborn, Plotly) and BI tools (Tableau/Power BI/Looker)
  • MLOps & version control — Git, MLflow or equivalent, Docker for reproducible model packaging

How you work

  • Relentlessly questions whether a correlation is truly causal before recommending spend reallocation — digs into confounders and selection bias.
  • Explores unconventional data sources (geolocation, weather, event calendars, competitor signals) to extract marginal predictive value for ROI models.
  • Continuously challenges the status quo of traditional attribution by prototyping privacy-preserving measurement techniques and LLM-augmented insights.

Skills we check

  • Causal inference & uplift modeling
  • Time-series forecasting & feature engineering
  • Bayesian & probabilistic modeling
  • Python (scikit-learn, XGBoost, Prophet, PyMC)
  • Marketing mix modeling (MMM) & multi-touch attribution
  • SQL & cloud data platforms (BigQuery/Snowflake)
  • Experiment design & A/B testing
  • Data visualization & stakeholder storytelling

What you get

  • Competitive annual compensation (₹12L–₹20L) with performance-linked bonuses and early equity eligibility.
  • Fully remote setup with a home-office stipend and flexible hybrid days at our Bengaluru headquarters.
  • Access to dedicated GPU credits and premium cloud compute for model training and experimentation.
  • Structured mentorship from senior data scientists and marketing-science leads, plus annual conference budget for NeurIPS, WSDM, or INMA.
  • Health insurance covering self and family, generous leave policy, and a sabbatical program after 4 years of service.

How we select for this role

We read every application ourselves. Here is what happens after you apply.

  1. Apply with a short form

    Paste a link to your CV and answer five questions in your own words. No sign-in and no file upload. About fifteen minutes.

  2. Integrity check

    When you press Check my answers, an automated check confirms the answers were typed by you. You see the result straight away, before you submit.

  3. Choose your project date

    If no cheating is detected you are shortlisted, and we email you a link to choose the day your project starts.

  4. 5-day practical project

    On that day our team contacts you with the brief: real work, 5 working days, reviewed by an engineer. For this role: Over a 5-working-day evaluation milestone, candidates will receive an anonymized, realistic dataset spanning 18 months of campaign-level ad-spend, impressions, clicks, conversions, and revenue across 6 digital channels and 4 geographies. Deliverables include: (Day 1–2) exploratory data analysis and feature-engineering blueprint with a diagnostic report; (Day 3–4) a calibrated marketing mix model (Bayesian or frequentist) that produces channel-level ROI estimates with credible/confidence intervals, plus sensitivity analysis on media-efficiency saturation curves; (Day 5) an executive-facing presentation and a short technical appendix explaining assumptions, validation strategy (back-testing & holdout metrics), and one recommended budget reallocation with expected ROI delta. Evaluation focuses on analytical rigor, reproducibility, clear communication, and practical business impact.

  5. Offer

    If your work solves the problem well, you receive your offer letter by email and in the portal.