Data Scientist

WTP Cost Management; Mumbai; On-site; INR 1,200,000 - 1,900,000; Full time

Job summary

WTP Cost Management is seeking a data-oriented professional in Mumbai to lead data extraction, cleaning, and transformation across diverse sources such as Excel, PDFs, and engineering documents. The role focuses on standardising datasets for analysis and enabling automation with ML/LLM techniques.

You will build templates, pipelines, and guardrails, collaborating with domain experts to translate cost and procurement data into actionable insights using regression, time-series, and NLP methods.

Qualifications

• 3-7 years of relevant experience in data science, data engineering, or cost/operations data domains.
• Background in SCM, procurement, logistics, construction/engineering, or industrial/manufacturing datasets.

Responsibilities

• Build automated workflows to extract data from Excel, PDFs, BOQs, engineering docs, emails, and semi-structured files.
• Apply OCR, document intelligence, and LLM-based extraction for non-standard formats.
• Create resilient pipelines for noisy, incomplete, and variable datasets.
• Define rules for data consistency, deduplication, and unit standardisation.
• Normalise data across regions and business units into analysis-ready datasets.
• Develop templates for mapping, classification, and feature engineering; automate multi-source outputs.
• Leverage LLMs to interpret documents, map data, and build prompts and guardrails.
• Develop models across regression, time-series, clustering, and NLP; perform feature engineering and tuning.
• Build end-to-end workflows and production-grade code with CI/CD integration.
• Maintain documentation for schemas, extraction logic, models, prompts, and validation rules.
• Collaborate with Digital & Data, domain experts, and IT infra teams to translate business logic into robust data/ML implementations.

Skills

Data extraction

Data cleaning

ML & forecasting

NLP

Education

Bachelors/Masters in Data Science, CS, Statistics, Engineering, Mathematics

Job description

Role & responsibilities

• Raw Data Extraction & Parsing
  • Build automated workflows to extract data from Excel (multi-tab, inconsistent formats), PDFs (native & scanned), BOQs, engineering docs, emails, and semistructured project files.
  • Apply OCR, document intelligence, and LLM-based extraction to interpret non-standard formats.
  • Create resilient pipelines for noisy, incomplete, and variable datasets.
• Data Cleaning, Normalisation & Standardisation
  • Define rules for data consistency, missing values, deduplication, units/convention standardisation.
  • Normalise data across regions, sectors, and business units.
  • Convert raw inputs into validated, analysis-ready datasets.
• Data Transformation & Template Building
  • Build summarised, structured templates (mapping rules, classification logic, grouping/segmentation, feature engineering).
  • Automate generation of consistent output from multi-source inputs; evolve templates over time.
• LLM & Prompt Engineering
  • Use LLMs to interpret complex documents, extract insights, classify and map data.
  • Develop prompt libraries, reusable chains, and guardrails for accuracy/reproducibility.
• Machine Learning Model Development
  • Develop models across regression, boosting, time-series forecasting, clustering/similarity, and NLP where relevant.
  • Perform feature engineering, hyperparameter tuning, crossvalidation, and model interpretation.
• Pipelines, Automation & MLOps Awareness
  • Build end-to-end workflows (extraction cleaning standardisation transformation modelling).
  • Write modular, productionready code and integrate with CI/CD where applicable.
• Documentation & Governance
  • Maintain documentation for schemas, extraction logic, models, prompts, and validation rules.
  • Ensure reproducibility, version control, and code quality (PEP8 or equivalent).
• Collaboration
  • Work with Digital & Data, domain experts (costing/engineering/PM), and IT infra teams.
  • Translate domain rules and business logic into robust data/ML implementations.

Preferred candidate profile

• Background in SCM, procurement, logistics, construction/engineering, or industrial/manufacturing datasets.
• Comfort dealing with regional variance and non-standard data schemas.
• Bachelors/Masters in Data Science, Computer Science, Statistics, Engineering, Mathematics, or related fields.
• 3-7 years of relevant experience (flexible for exceptional talent).