Data Scientist

Hinduja Global Solutions (HGS) is a global company specializing in business process management (BPM) and digital customer experience (CX) solutions. HGS offers a range of services including contact center solutions, back-office processing, and integrated digital platforms. They are part of the Hinduja Group, a large multinational conglomerate. With over 20,000 employees across 34 delivery centers in 8 countries, HGS offers a spectrum of technology, digital marketing, and outsourcing solutions tailored to specific industry verticals.

HGS is looking for someone with the following -

Core Skills & Knowledge

• Programming: Expert Python; deep use of Pandas, NumPy, Scikit-learn, PyTorch/Lightning; strong OO and functional patterns.
• Large Language Models: Fine-tuning (LoRA, QLoRA, PEFT), prompt engineering, evaluation (BLEU, BERTScore, hallucination metrics).
• Machine Learning: Classical algorithms (tree-based, linear, ensemble, time-series), hyper-parameter search (Optuna, Ray Tune), model explainability (SHAP, Captum).
• Statistical Methods: Experimental design, p-value and power analysis, Bayesian A/B testing, causal inference.
• Vector Search & RAG: Embedding generation, similarity search tuning (HNSW, IVF-PQ), chunking strategies, knowledge-base hygiene.
• DevOps & Tooling: Git, Docker, Conda/Poetry, make/Task, notebook versioning, automated linting & unit tests (pytest).
• Cloud & Data Platforms: Familiarity with AWS (SageMaker, Bedrock) or Azure (ML, OpenAI) plus modern data warehouses (Snowflake, Redshift, BigQuery) and lake formats (Parquet, Delta).

Why this role matters:

We deliver data-driven and GenAI-powered products that move business KPIs, not just models. Youll transform raw data and cutting-edge research (LLMs, causal inference, RAG pipelines) into production-grade capabilities that customers use every day.

What you'll do:

Model development & experimentation – Build and validate supervised/unsupervised models (classification, regression, clustering) and LLM fine-tuning pipelines; design statistically rigorous experiments (A/B, multivariate, uplift). This also includes RAG applications, and tune RAG workflows with vector databases (pgvector, Milvus) and embeddings (OpenAI, Hugging Face). Optimize retrieval recall, latency, and hallucination rates.

Causal & statistical analysis – Apply quasi-experimental methods (DID, propensity scoring, synthetic controls) and hypothesis tests to uncover drivers of business outcomes.

Operationalization & MLOps – Package models as reproducible artifacts (MLflow, TorchScript, BentoML), write CI/CD tests, and partner with MLOps engineers on scalable deployment and monitoring.

Data wrangling & feature engineering – Build robust ETL/ELT pipelines in Python (Pandas, Polars, SQL) and orchestrate workflows (Airflow, Dagster).

Collaboration & knowledge sharing – Translate findings to non-technical stakeholders, contribute to shared notebooks, mentor junior analysts, and document best practices.

The ideal applicant will have the following experience & education

• 6-10 years of overall IT experience with over 5 years of relevant experience building and shipping ML models or GenAI services in production.
• Proven track record of improving a business metric (conversion, churn, NPS, cost) through data science.
• Bachelor’s or Master’s in Computer Science, Statistics, Mathematics, or related field (or equivalent hands-on experience).
• English fluency

Why HGS?

At HGS, we offer our employees get an open and collaborative environment and gain experience working for customers in various industries while solving complex business problems. They get support to learn new technologies as well as to enhance existing skills to further their career growth with competitive remuneration, excellent performance based annual increments, Fast paced growth opportunities, International work experience.