4492 - Lead Data Scientist

· Architecture. Design production agentic and RAG systems that meet real customer scale and reliability requirements, not benchmark conditions.

· Orchestration depth. Apply agent design patterns with judgment — memory, tool routing, multi-agent coordination, and explicit failure handling — and decide deliberately where autonomy stops and a human takes over.

· Retrieval quality. Own the retrieval pipeline as a first-class system: chunking strategy, embedding model selection, vector stores, re-ranking, and relevance tuning against measured outcomes.

· LLMOps lifecycle. Own model and prompt versioning, eval pipelines running in CI, observability (tracing, token and cost dashboards), and the guardrails and safety filters that ship with every release.

· Model strategy. Select, fine-tune, and serve SLMs and open models — LoRA and QLoRA, quantization, inference optimization, GPU and serving trade-offs.

· Engineering economics. Optimize latency, cost, and accuracy together, and make defensible build-versus-buy and model-selection calls you can explain to both engineers and executives.

· Define and execute the quarterly roadmap for your area, and translate ambiguous business problems into machine learning problems with clear solution workflows.

· Work with business leaders and customers directly to understand where the workflow actually breaks, then build for that.

· Partner with the data platform and applications teams so your capabilities land inside their products and workflows rather than beside them.

· Set the coding and evaluation standards for your team, and lead design reviews.

· Pursue published work or patents where the problem warrants it — particularly in healthcare AI.

· 6+ years in data science, applied ML, or AI engineering, including 2+ years building LLM-powered products. Healthcare experience is a plus.

· Deep NLP and GenAI experience. Statistical and classical machine learning is good to have on top of that.

· Strong hands-on Python — building highly scalable, performant enterprise applications, plus optimization technique.

· Hands-on experience with deep learning frameworks: PyTorch and/or HuggingFace transformers.

· At least one shipped GenAI product with a genuinely complex architecture — multiple agents, memory, retrieval, and agent OTEL/tracing in production.

· Working command of modern fine-tuning: PEFT methods, with LoRA and QLoRA preferred.

· Hands-on experience with at least one ML platform — Databricks, Azure ML, or SageMaker.

· Experience leading engineers, formally or as a technical lead — you have owned other people’s output, not only your own.

· Strong written and spoken communication, with a customer-focused instinct in both conversation and documentation.

· Preferably a Master’s in Computer Science, Computer Engineering, or a related field.

Everything above sits on top of independent delivery. This role assumes you can already do the following without supervision:

· Build production-grade RAG and LLM/SLM-powered features end to end with limited supervision.

· Work fluently in at least one orchestration framework — LangGraph, LlamaIndex, CrewAI, or equivalent — to compose multi-step, tool-using flows.

· Design retrieval pipelines and tune them for relevance: chunking, embeddings, vector stores, re-ranking.

· Implement prompt engineering, function and tool calling, and reliable structured-output parsing.

· Write and run evals — golden sets, LLM-as-judge — to measure quality and catch regressions before customers do.

· Containerize and deploy services (Docker, REST/gRPC) with an eye on latency, token cost, and basic guardrails.

· Document well and participate actively in code review.

· API frameworks for robust web applications — FastAPI or Django preferred.

· Comfort with at least one hyperscaler cloud.

· Papers or patents, especially in healthcare AI.

We offer competitive benefits to set you up for success in and outside of work.

Our Noida office is situated in a posh techspace, equipped with various amenities to support our work environment. Here, we follow a five-day work schedule, allowing us to efficiently carry out our tasks and collaborate effectively within our team.

Innovaccer is an equal-opportunity employer. We celebrate diversity, and we are committed to fostering an inclusive and diverse workplace where all employees, regardless of race, color, religion, gender, gender identity or expression, sexual orientation, national origin, genetics, disability, age, marital status, or veteran status, feel valued and empowered.

Disclaimer: Innovaccer does not charge fees or require payment from individuals or agencies for securing employment with us. We do not guarantee job spots or engage in any financial transactions related to employment. If you encounter any posts or requests asking for payment or personal information, we strongly advise you to report them immediately to our Px department at px@innovaccer.com. Additionally, please exercise caution and verify the authenticity of any requests before disclosing personal and confidential information, including bank account details.

Innovaccer Inc. is the data platform that accelerates innovation. The Innovaccer platform unifies patient data across systems and care settings, and empowers healthcare organizations with scalable, modern applications that improve clinical, financial, operational, and experiential outcomes. Innovaccer’s EPx-agnostic solutions have been deployed across more than 1,600 hospitals and clinics in the US, enabling care delivery transformation for more than 96,000 clinicians, and helping providers work collaboratively with payers and life sciences companies. Innovaccer has helped its customers unify health records for more than 54 million people and generate over $1.5 billion in cumulative cost savings. The Innovaccer platform is the #1 rated Best-in-KLAS data and analytics platform by KLAS, and the #1 rated population health technology platform by Black Book. For more information, please visit innovaccer.com .

Check us out on YouTube , Glassdoor , LinkedIn , and innovaccer.com .