Data Scientist

Employment Type

About the Role

• Own end-to-end delivery of impactful ML/NLP/ GenAI solutionsfrom problem framing and data strategy to modeling, deployment, and monitoring. Youll lead scoped projects, collaborate with product/engineering, and help raise the bar on model quality, safety, and reliability.

What Youll Do

Problem Framing & Impact

• Translate ambiguous requirements into clear ML objectives with measurable success metrics, latency/cost constraints, and guardrails.
• Modeling & Algorithms
• Build/evaluate models (classification, ranking, sequence models, embeddings, LLMs; RAG/agents).
• Run error analysis, ablations, and A/B tests; iterate quickly.

GenAI

• Build LLM apps with LangChain / LangGraph (prompting, tools/agents, memory, evaluators). Implement RAG with vector stores and add guardrails/evaluation.

Data & Features

• Partner on data sourcing/labeling, feature & prompt engineering; ensure data quality SLAs and lineage.

Production &MLOps

• Ship models as APIs/batch jobs ( FastAPI /Flask, Docker). Use CI/CD, model registries ( MLflow /W&B), and implement monitoring for drift/quality/latency/cost.

Collaboration & Communication

• Write design docs, present insights to technical/non-technical audiences, and mentor interns/juniors on best practices.

Minimum Qualifications

• Bachelors/Masters in a quantitative field (CS, Data Science, Math, Stats, EE) and 2-5 years hands-on ML experience.
• Strong Python and SQL; solid statistics, experimental design, and data wrangling.
• Proven delivery of production ML/DL systems (scikit-learn, PyTorch or TensorFlow).
• Practical NLP with Transformers/Hugging Face; embeddings and text classification/generation.
• Experience building LLM apps with LangChain / LangGraph and vector databases (FAISS/Milvus/Pinecone).
• Production experience with APIs ( FastAPI /Flask), containers (Docker), orchestration (Airflow/Prefect), and CI/CD on cloud (AWS/GCP/Azure).
• Clear written & verbal communication; ability to lead scoped projects end-to-end.

Preferred (Nice to Have)

• Data-at-scale tools (Spark/ PySpark or Polars); streaming exposure (Kafka/Pulsar) is a plus.
• ML platforms: feature stores, model serving (Triton/ SageMaker /Vertex), and evaluation frameworks.
• Governance/lineage ( DataHub / OpenLineage ), privacy/PII handling, and security best practices.
• Publications, OSS contributions, or notable internal tech write-ups.
• Tools & Tech You May Use
• Python, SQL, scikit-learn, PyTorch /TensorFlow, Hugging Face, LangChain , LangGraph , vector stores (FAISS/Milvus/Pinecone), FastAPI , Docker, MLflow /W&B, Airflow/Prefect, cloud (AWS/GCP/Azure), Prometheus/Grafana/Datadog.