Senior Data Scientist
On-sitePune, Maharashtra, India
Job Summary
Design and build end-to-end GenAI applications, including RAG systems, agentic workflows, and multi-agent orchestration. Own architecture decisions across model selection, retrieval design, and serving while establishing evaluation frameworks and guardrails for production reliability. Partner with stakeholders to translate ambiguous problems into scoped, high-ROI AI solutions and set technical standards by reviewing designs and mentoring mid-level engineers. Stay current with the field to bring practical recommendations on prompt engineering, fine-tuning, and agentic system frameworks like LangChain or AutoGen. Requires 8–12 years of data science experience with production-deployed models and strong Python, SQL, and PyTorch skills.
Required Qualifications
- 8–12 years in data science / ML, with a track record of production-deployed models and systems
- Strong foundations in ML and statistics: supervised/unsupervised learning, experimentation and A/B testing, model evaluation; familiarity with causal inference and time series
- Expert Python (pandas, NumPy, scikit-learn), strong SQL, and proficiency in PyTorch or an equivalent DL framework
- Demonstrated experience building GenAI applications: prompt engineering, RAG (chunking, embeddings, vector databases, reranking), and fine-tuning/adaptation (SFT, LoRA/PEFT) with sound judgment on when to use each
- Hands-on experience designing agentic systems: tool/function calling, planning and orchestration, memory and state management, using frameworks such as LangChain/LangGraph, LlamaIndex, CrewAI, or AutoGen (and knowing when to build directly)
- Experience with LLM evaluation and observability (e.g., LLM-as-judge, golden datasets, tracing tools like LangSmith/Langfuse/Arize) and with guardrails for hallucination, safety, and prompt-injection defense
- Production engineering skills: API design, serving and streaming, caching/batching, rate-limit handling, Docker, and CI/CD
- Strong communication skills and experience mentoring or leading technically
Desired Qualifications
- Inference optimization experience (quantization, vLLM/TGI, GPU serving) and cost optimization at scale
- Familiarity with MCP and emerging model-to-system interoperability standards
- Kubernetes and infrastructure-as-code (Terraform)
- Experience with multimodal or unstructured data pipelines
Hiring someone like this?
Get your role in front of qualified candidates on Sorce.