Key Responsibilities

Reasoning & Requirement Refinement

- Challenge unclear or technically weak requirements with practical alternatives.
- Translate clinical and product requirements into clear, testable technical specifications.
- Proactively identify risks and failure modes before implementation.

GenAI & Conversational AI (80%)

- Design and build end-to-end RAG pipelines (ingestion, chunking, embeddings, vector stores, retrieval, generation).
- Integrate and evaluate LLMs (OpenAI, Claude, Gemini) with a focus on response quality, hallucination mitigation, and healthcare safety.
- Develop AI services using LangChain/LangGraph, Python, Flask, and REST APIs.
- Build prompt engineering and LLM evaluation frameworks covering relevance, accuracy, safety, and tone.
- Implement and optimize vector databases (FAISS, Pinecone, Weaviate) and embedding pipelines.

Machine Learning & Signal Engineering (1020%)

- Develop ML models and health signals from physiological data (CGM, HRV, sleep, activity, heart rate).
- Engineer meaningful features for time-series health data and evaluate model confidence, accuracy, and edge cases.
- Apply explainable and clinically defensible ML approaches, choosing the simplest effective model.

Data Engineering

- Build, debug, and maintain data pipelines in Python and C#.
- Manage end-to-end health data flow from connected devices through ingestion, transformation, storage, and AI consumption.
- Resolve data quality issues and work with structured/time-series data, including normalization, windowing, gap handling, sensor dropouts, and timezone-aware aggregation.





Healthcare AI & Production

- Build clinically safe AI systems that detect and surface incorrect or uncertain outputs.
- Ensure HIPAA-compliant data handling and healthcare best practices.
- Collaborate closely with clinicians to develop production-ready AI features.
- Own features end-to-end, from requirements and deployment to monitoring, evaluation, and production debugging.
- - Non-negotiable
- Demonstrated reasoning ability the capacity to look at an ambiguous problem, break it down, identify what is
- missing, and propose a path forward. This is the primary screen.
- - Genuine hands-on GenAI experience — building RAG pipelines, LLM-integrated features, prompt engineering
- with evaluation. Real systems, real failure modes, real fixes. Not course projects.
- Python proficiency — clean, testable, production-quality code. Flask or equivalent API framework experience.
- C# / .NET literacy — you can read, debug, and contribute to data pipelines written in C#. You do not need to be
- a C# architect, but data engineering is a daily reality in this role and you engage with it directly.
- Working knowledge of LLM integration: prompt engineering, evaluation, hallucination risk, and safety
- constraints.



You have a mental model of when models fail and why.
- Enough ML grounding to think clearly about signals from health data — HRV, CGM, sleep, activity — without
- needing to be a deep learning researcher.
- Vector database experience — you understand retrieval quality trade-offs, not just the API calls.
- Ownership instinct — the kind of person who is bothered by a bug in production even when it isn't technically
- assigned to them.
- Curiosity that is visible — in side projects, questions you ask, papers you've read, things you've broken on
- purpose to understand them.
- Advantageous:
- Strong advantage
- Experience building ML-based health signals from physiological time-series data — HRV, sleep stages, CGM,
- activity — and a clear understanding of what makes those signals reliable or unreliable.
- Experience in healthcare or health-tech AI — clinical accuracy requirements, secure AI design, and the stakes of a
- wrong answer in a medical context.
- LangChain or LangGraph experience — particularly building multi-step pipelines with memory, tool use, and
- evaluation.
- Cloud platform experience (AWS, GCP, or Azure) — deploying and monitoring AI services in production.
- GraphQL APIs and Next.js — for collaboration with the frontend team.
- Familiarity with HIPAA-compliant data pipelines and healthcare data handling standards.
- Experience building AI evaluation frameworks — not just shipping models, but measuring whether they are
- working.

📌 AI / ML Engineer - GenAI - Data (Mumbai)
🏢 Continuity 1
📍 Mumbai