Data Scientist ONSITE- Bengaluru

Uplers Solutions Private Limited.; Bengaluru; On-site; INR 3,500,000 - 4,000,000; Full time

Job summary

Uplers Solutions Private Limited in Bengaluru is seeking a hands-on Data Scientist with 3+ years of experience to build production-grade models for customer 360, eligibility, and activation workflows on AWS.

You will work across data science, analytics, engineering, risk, marketing technology, and business execution, translating business problems into scalable features and reliable scoring services using Python, PySpark, SQL and ML tooling.

Qualifications

• 3–6 years of experience in data science, analytics or fintech.
• Proficiency in Python and SQL.
• Experience with Airflow, PySpark and AWS.
• Familiarity with MLflow or SageMaker for production ML.

Responsibilities

• Build supervised ML models for propensity, conversion, eligibility, affinity, next-best-action and lead prioritization problems.
• Define labels, horizons and leakage-safe feature cutoffs for financial services use cases.
• Evaluate models using AUCPR, precision@K, recall@K, lift@K, calibration and business impact.
• Create ranked score outputs for CRM, CleverTap, call center and product teams.
• Collaborate with Data Engineering to productionize training, scoring and activation pipelines on AWS.

Skills

Python

SQL

Machine Learning

Airflow

PySpark

AWS

Model Validation

Feature Stores

Tools

Airflow

PySpark

AWS

MLflow

SageMaker

SQL

Job description

Job Description:

Experience: 3.00 + years

Salary: INR 3500000-4000000 / year (based on experience)

Expected Notice Period: 7 Days

Shift: (GMT+05:30) Asia/Kolkata (IST)

Opportunity Type: Office ()

Placement Type: Full Time Permanent position(Payroll and Compliance to be managed by: Leading Financial App)

(*Note: This is a requirement for one of Uplers' client - Leading Financial App)

What do you need for this opportunity?

Airflow, Feature Stores, ML Flow, SQL, Python, PySpark, AWS, Model Validation

Role Summary

We are hiring a hands-on Data Scientist to build production-grade models and decisioning intelligence across Custome 360, cross-sell targeting, lead prioritization, bureau-driven eligibility, campaign effectiveness measurement, and real-time customer activation. The role sits at the intersection of data science, analytics, engineering, risk, marketing technology, and business execution.

About The Function

• Customer 360, bureau and app-event feature engineering, eligibility and affinity models, lead prioritization, cross-sell targeting, campaign measurement, analytics governance, and production data pipelines on AWS.
• Customer 360 and Next Best Action: build unified customer profiles from MFLO transactions, app behavior, CRIF/CIBIL bureau data, lead journeys and campaign interactions.
• Eligibility and Affinity Engine: combine rule-based eligibility, bureau-derived features and ML models to identify the right product, right customer and right activation route.
• Activation and measurement: push model outputs to CleverTap, call-center workflows, CRM and dashboards while measuring incremental uplift through holdouts, experiments and attribution logic.
• Production orientation: models must move beyond notebooks into scheduled pipelines, tracked experiments, auditable outputs and reliable scoring services.

Role Purpose

The Data Scientist will own model development and analytical experimentation for customer targeting and conversion use cases. This person should be able to translate business problems into modelling tasks, build reliable features from large- scale financial and behavioral data, evaluate model lift properly, and work with engineering to deploy outputs into production workflows.

Key Projects the Candidate Will Work On

• Customer 360 feature intelligence: create customer-level features using transactions, app journeys, lead history, bureau attributes, campaign exposure and engagement signals.
• PL / CC / Gold Loan / Digital Gold propensity models: build product-specific affinity and conversion models for cross- sell, lead prioritization and campaign targeting.
• Eligibility + BRE parity modelling: use scrub/bureau data to approximate rule-engine decisions, analyze false pass/fail cases, and improve eligibility pre-screening.
• Lead intent and conversion modelling: develop separate top-of-funnel intent and post-lead conversion models, with rule-engine gates and product-specific constraints.
• CleverTap and campaign effectiveness analytics: design experiments, holdouts, attribution windows and campaign decision tables to measure uplift and prevent audience overlap.
• Real-time and batch scoring: support SageMaker scoring, S3 score snapshots, Glue feature pipelines, DynamoDB hot features and ECS/Fargate scoring services.
• Model observability and governance: track model variants, feature sets, validation metrics, lift@K, precision@K, drift checks, suspicious feature reports and run metadata using MLflow or equivalent tracking.

Responsibilities

Model Development and Evaluation

• Build supervised ML models for propensity, conversion, eligibility, affinity, next-best-action and lead prioritization problems.
• Define appropriate labels, horizons, observation windows and leakage-safe feature cutoffs for financial services use cases.
• Evaluate models using AUCPR, precision@K, recall@K, lift@K, calibration, cohort-level performance and business conversion impact.
• Create candidate lists and ranked score outputs that can be used by CRM, CleverTap, call center, product and revenue teams.
• Compare model variants such as bureau-only, behavior-only, lead-gated, rule-gated and hybrid models.

Feature Engineering and Data Understanding

• Engineer features from bureau data, app events, AppsFlyer/CleverTap events, transaction history, lead tables, call- center outcomes and product journeys.
• Work with time-aware snapshots and avoid leakage in joins, target creation and validation windows.
• Create reusable feature logic in PySpark/Python that can run on AWS Glue and be consumed by SageMaker or downstream scoring jobs.
• Perform deep-dive analysis on data quality, null coverage, distribution shifts, event instrumentation gaps and identity matching issues.

Experimentation, Measurement and Business Analytics

• Design A/B tests, holdouts, uplift measurement and attribution frameworks for campaigns and targeting programs.
• Partner with Marketing, Revenue, Product and Call Center teams to translate model scores into measurable business actions.
• Create clear analysis explaining why a cohort was selected, how campaign performance should be measured and what decisions should change.
• Support dashboards and MIS for funnel tracking, campaign performance, lead-to-conversion, disbursal/GPV and channel performance.

Productionization and Collaboration

• Collaborate with Data Engineering to productionize training, scoring and activation pipelines using S3, Glue, Redshift/Athena and SageMaker.
• Log experiments, model metadata, feature lists, configuration, validation metrics and artifacts for reproducibility.
• Help define monitoring for model drift, score distribution changes, pipeline failures and activation quality.
• Document modelling assumptions, definitions, guardrails and business interpretation in a way non-technical stakeholders can understand.

Technology Stack

Core ML / Analytics: Python, pandas, scikit-learn, XGBoost/LightGBM, statistical analysis, model evaluation, experimentation

Big Data / ETL: PySpark, AWS Glue, SQL, partitioned S3 datasets, parquet/iceberg-style data organization

Warehouse / Query: Amazon Redshift, Athena, SQL optimization, data marts and analytical tables

Cloud / ML Ops: AWS S3, SageMaker training/scoring, MLflow or equivalent experiment tracking, CloudWatch basics

Serving / Activation: Batch scoring, score snapshots, DynamoDB hot feature lookup, ECS/Fargate scoring service awareness

MarTech / Product Analytics: CleverTap, AppsFlyer, event funnels, campaign exposure/click/conversion analytics, cohort activation

Visualization / Reporting: QuickSight, Power BI, Tableau or equivalent dashboarding exposure

Ways of Working: Git/Bitbucket, code reviews, reproducible notebooks/scripts, documentation and stakeholder communication

Required Skills And Experience

• 3–6 years of experience in data science, applied machine learning, risk analytics, marketing analytics, customer analytics or fintech analytics.
• Strong Python and SQL skills, with proven ability to work on large datasets and messy real-world business data.
• Experience building classification, ranking, propensity, conversion, churn, cross-sell, credit/risk or recommendation models.
• Strong understanding of model validation, time-based splits, leakage prevention, class imbalance, lift charts and business-focused evaluation.
• Ability to explain model outputs, trade-offs and limitations clearly to business, product and engineering stakeholders.
• Comfort working in a cloud data environment, ideally AWS-based, with S3, Glue, Redshift/Athena and SageMaker exposure.
• Good documentation discipline: feature definitions, assumptions, run metadata, experiment notes and model interpretation

Good to Have

• Financial services, lending, bureau data, credit risk, fintech, NBFC or banking analytics experience.
• Experience with CIBIL/CRIF-style bureau attributes, eligibility rules, underwriting policy analytics or BRE systems.
• Exposure to CleverTap, AppsFlyer, campaign analytics, customer journey events and marketing attribution.
• Experience with MLflow, model registries, feature stores, Airflow/Step Functions/EventBridge or production ML monitoring.
• Ability to work with PySpark at scale and convert notebook logic into reliable production jobs.
• Experience designing holdout tests, uplift measurement and incrementality studies for CRM/campaign programs.

Success Metrics for This Role (First 90–180 Days)

• Model pipeline contribution: Contributes to at least one production-grade affinity / conversion / eligibility model with documented features, validation and scoring output.
• Business activation: Creates ranked cohorts that are usable by CleverTap, CRM, call center or product teams with clear eligibility and targeting logic.
• Measurement discipline: Defines holdout or experiment design and reports campaign/model uplift using agreed metrics.
• Engineering collaboration: Works with DE to make feature generation and scoring repeatable on AWS instead of notebook-only execution.
• Documentation and stakeholder trust: Produces crisp documentation that explains model objective, data, label, validation, limitations and interpretation.

Ideal Candidate Profile

• A builder who can start from an ambiguous business problem and convert it into data, labels, features, experiments and production-ready outputs.
• Strong enough technically to write reliable Python/SQL/PySpark, but business-oriented enough to care about conversion, eligibility, revenue impact and operational adoption.
• Comfortable with imperfect data, changing definitions, data discovery and stakeholder-driven prioritization.
• Bias for simple, explainable and measurable solutions before adding modelling complexity.
• Able to collaborate with Data Engineering, Product, Risk, Revenue, CRM, Marketing and Call Center teams without waiting for perfect requirements.