FirstHive - Data Science Engineer

FirstHive | CDP+AI Data Platform; Bengaluru; On-site; INR 2,400,000 - 3,800,000; Full time

Job summary

FirstHive | CDP+AI Data Platform in Bengaluru seeks a Data Science Engineer to build and deploy production ML models and AI features for its CDP platform. You will work in a compact AI & Data Science team on customer segmentation, entity resolution, predictive analytics, NLP, and LLM-powered automation that drives how enterprise clients engage with customers.

The role is hands-on, shipping models to production, not notebooks.

Qualifications

• Proficient in Python ML stack (scikit-learn, Pandas, NumPy, XGBoost) for production models.
• Experience building and deploying ML models in production with APIs (FastAPI/Flask) and containerization.
• Strong SQL skills for feature engineering on Snowflake/BigQuery.
• Experience with NLP, embeddings, and LLM-based features.
• Experience with entity resolution and data integration.
• Familiarity with MLOps concepts.

Responsibilities

• Build and deploy customer segmentation and clustering models at scale.
• Develop entity resolution algorithms and probabilistic matching across data sources.
• Build predictive models (churn, conversion propensity) using classification and regression.
• Design LLM-powered features and AI-driven insights with prompt engineering and RAG pipelines.
• Develop NLP capabilities: embeddings, entity extraction, text classification.
• Write complex SQL for feature engineering on Snowflake/BigQuery.
• Integrate ML models into core platform via APIs (FastAPI) for real-time and batch inference.
• Own model lifecycle: monitoring, drift detection, retraining, versioning.

Skills

Python ML stack

LLM / GenAI

SQL

NLP

Model deployment

Entity resolution

Tools

FastAPI

Docker

Snowflake

BigQuery

Pinecone

FAISS

Job description

Role Description

We are seeking a Data Science Engineer to build and deploy production ML models and AI features for our CDP platform. You will work in a small, high-ownership AI & Data Science team building customer segmentation models, entity resolution algorithms, predictive analytics, NLP capabilities, and LLM-powered automation that directly impact how enterprise clients understand and engage with their customers. This is a hands‑on engineering role you build models that ship to production, not notebooks that stay in research.

Designation : Data Science Engineer

Location : Bengaluru

Experience : 4 - 6 years

Function : AI & Data Science

Key Responsibilities

• Build and deploy customer segmentation and clustering models (K-Means, DBSCAN, hierarchical) at scale.
• Develop entity resolution algorithms fuzzy matching, blocking strategies, probabilistic scoring to unify customer profiles across disparate data sources.
• Build predictive models churn prediction, conversion propensity, next-best-action recommendations using classification and regression (XGBoost, Random Forest, logistic regression).
• Design and build LLM-powered features schema mapping automation, natural language querying, AI-driven insight generation using prompt engineering, RAG pipelines, and structured output extraction.
• Build NLP capabilities text embeddings, semantic similarity, entity extraction, text classification using transformers (BERT or similar).
• Write complex SQL for feature engineering window functions, sessionization, time-series aggregation, customer behavior features from raw event data on Snowflake/BigQuery.
• Integrate ML models into the core platform via APIs (FastAPI) for real‑time and batch inference.
• Own model lifecycle in production monitoring, drift detection, retraining, versioning.
• Work with data engineering teams to ensure clean, structured training data and feature pipelines.

Experience And Skills

• Python ML stack : scikit-learn, Pandas, NumPy, XGBoost. This is 70% of the work.
• LLM / GenAI : prompt engineering, RAG fundamentals, embeddings, vector similarity search, calling LLM APIs (Claude, OpenAI, or similar) with structured outputs. Not fine-tuning effective use of APIs.
• SQL : complex feature extraction queries on analytical databases. Window functions, sessionization, time-series aggregation, cost-aware query patterns. Not basic SELECT.
• NLP : text embeddings (sentence-transformers or similar), named entity recognition, text classification, semantic search.
• Model deployment : FastAPI or Flask, Docker containerization, REST API serving.
• Entity resolution / record linkage fuzzy matching (Levenshtein, Jaro‑Winkler), blocking strategies, probabilistic matching across multiple fields.
• Model evaluation precision/recall trade-offs, cross-validation, A/B testing.

Good To Have

• LangChain, vector databases (Pinecone, FAISS).
• MLflow or experiment tracking.
• Kafka consumers for real‑time scoring.
• Snowflake ML / BigQuery ML.
• Time-series forecasting (Prophet, ARIMA).
• Customer analytics or MarTech platform experience.

(ref:hirist.tech)