Job Description:
Senior Data Scientist — ML & Semantic AI
Technologies: Azure · NLP · RAG · Semantic Matching · Python
Role SummaryWe are looking for a Data Scientist with expertise in Python, Azure Cloud, and NLP to build and enhance machine learning models at scale. The role includes embedding optimisation, semantic matching, LDA and RAG architectures, dense and sparse retrieval pipelines, and migration of cloud-native data pipelines to Azure Databricks.
Core Requirements- Design and execute end-to-end machine learning pipelines including data extraction, preprocessing, feature engineering, model development, tuning, and deployment.
- Develop machine learning pipelines using Azure Synapse, Databricks, and Snowflake.
- Build and deploy classification, regression, and clustering models.
- Develop and deploy proof-of-concept solutions for client use cases.
- Implement semantic matching and similarity search using cosine similarity, dot-product scoring, and bi-encoder/cross-encoder architectures (e.g., SBERT, sentence-transformers).
- Build embedding models by fine-tuning pre-trained models and optimising embedding storage in vector databases such as Chroma DB, FAISS, and Azure AI Search.
- Train and optimise models for new data providers with dynamic input handling.
- Improve LDA model performance for large-scale topic modelling.
- Implement hybrid semantic search by combining dense and sparse retrieval methods.
- Optimise RAG architectures and retrieval QA systems for chatbot and recommendation performance.
- Enable semantic query understanding using intent classification and query expansion techniques.
- Develop forecasting models for marketing, demand prediction, and trend analysis.
- Apply NLP-based forecasting techniques using sentiment and external data.
- Use semantic similarity for audience intelligence, including zero-shot and few-shot classification techniques.
- Migrate data pipelines from Azure Synapse to Azure Databricks and retrain models accordingly.
- Optimise embedding storage and retrieval within Azure AI Search.
- Perform vector index tuning including HNSW optimisation and ANN benchmarking for production systems.
Python, Azure Databricks, Azure ML, Azure Synapse, Azure Blob Storage, Scikit-learn, NumPy, Pandas, Hugging Face, sentence-transformers, FAISS, Chroma DB, Azure AI Search, LangChain, TensorFlow, PyTorch, Statsmodels, Azure OpenAI.
Location:
DGS India - Mumbai - Thane Ashar IT ParkBrand:
MerkleTime Type:
Full timeContract Type:
PermanentSkills Required
- Expertise in Python
- Experience with Azure Cloud platform (Azure Databricks, Azure ML, Azure Synapse, Azure Blob Storage, Azure AI Search)
- Experience building end-to-end ML pipelines including data extraction, preprocessing, feature engineering, model development, tuning, and deployment
- Experience with Snowflake
- Experience in NLP, semantic matching, and similarity search (cosine similarity, dot-product, bi-encoder/cross-encoder architectures)
- Experience with embeddings: fine-tuning pre-trained models and optimising embedding storage (Chroma DB, FAISS, Azure AI Search)
- Experience with RAG architectures, retrieval-augmented generation, and retrieval QA systems
- Experience implementing hybrid dense and sparse retrieval pipelines and vector index tuning (HNSW, ANN benchmarking)
- Experience with sentence-transformers / SBERT and Hugging Face ecosystem
- Experience building and deploying classification, regression, clustering, and forecasting models (including NLP-based forecasting)
- Proficiency with ML libraries: Scikit-learn, NumPy, Pandas, Statsmodels
- Experience with deep learning frameworks: TensorFlow and/or PyTorch
- Experience with LangChain and Azure OpenAI
- Experience migrating cloud-native data pipelines (Azure Synapse to Azure Databricks) and retraining models
dentsu Compensation & Benefits Highlights
The following summarizes recurring compensation and benefits themes identified from responses generated by popular LLMs to common candidate questions about dentsu and has not been reviewed or approved by dentsu.
-
Parental & Family Support — Paid parental leave at full pay and caregiver supports (including backup care) are emphasized as standout elements. Feedback suggests family-oriented benefits are a strong part of the package.
-
Leave & Time Off Breadth — Flexible or unlimited PTO, extensive paid holidays, and a year-end office closure are established components. Feedback suggests time-off policies are generous and add meaningful flexibility.
-
Retirement Support — A large, established 401(k) plan with employer matching is clearly documented. Feedback suggests retirement benefits feel competitive and straightforward.
dentsu Insights
What We Do
We are dentsu. We team together to help brands predict and plan for disruptive future opportunities and create new paths to growth in the sustainable economy. We know people better than anyone else and we use those insights to connect brand, content, commerce and experience, underpinned by modern creativity. We are the network designed for what’s next
