Health Research Engineering Intern

About the role

IAIRO is conducting applied research in health, biomedical AI, and personalized wellness. As a Health Research Engineering Intern, you will study relevant papers, summarize and compare methods, identify ideas that can be converted into practical product workflows, and help build lightweight prototypes. The role is suited to students and early-career researchers or engineers who can connect academic research with real-world MVP development while working responsibly with health information.

Responsibilities

• Read and summarize research papers related to AI hospital simulations, medical agents, personal health insights, Personicle, Personal Care Utility, medical reasoning, and responsible health AI.

• Convert academic ideas into product concepts, user workflows, architecture diagrams, evaluation plans, and MVP requirements.

• Research how multi-agent health systems work, including patient, doctor, nurse, examiner, health-guidance, and supervisory agents.

• Explore how wearable, lifestyle, symptom, sleep, activity, food, medication, and other health data can be combined to generate contextual and personalized insights.

• Design and run experiments with synthetic or de-identified data, document assumptions, and evaluate outputs against technically and clinically meaningful criteria.

• Build or support products using Python, APIs, notebooks, Streamlit, Gradio, or similar tools.

• Use LLMs, fine tune SLMs, pre-train and train small models to test agentic workflows, retrieval, code-generation, and structured reasoning approaches.

• Collaboration with clinicians, public-health researchers, engineers, and product leads to frame problems and validate outputs.

• Create structured research notes, comparison tables, benchmark summaries, architecture diagrams, and MVP documentation.

• Identify open-source tools, datasets, APIs, models, and frameworks that can accelerate responsible MVP development.

Research domains

• AI Hospital: Doctor-patient-examiner-chief physician simulations used to study and evaluate LLM behavior in clinical scenarios.

• Agent Hospital: Hospital-like multi-agent environments where patient, nurse, and doctor agents interact across simulated treatment cycles.

• PHIA / Personal Health Insights Agent: LLM-agent, retrieval, and code-generation approaches for analyzing wearable data and producing personalized insights.

• Personicle: Longitudinal representation of a person's daily activities, health events, behaviors, and context as a usable personal timeline.

• Personal Care Utility / PCU: Always-on personalized guidance using multimodal data, personal baselines, contextual reasoning, and continuous learning.

• Medical reasoning and chronic-care guidance: Doctor-style reasoning systems and practical lifestyle support for diabetes and other long-term conditions, with clear safety boundaries.

Required qualifications

• Currently pursuing a M.Tech / PhD (or equivalent) in Computer Science, Software Engineering, AI/ML, Data Science, Biomedical Informatics, or a related field.

• Ability to read, understand, critique, and clearly summarize technical research papers.

• Python programming and comfort working in Jupyter Notebook or a similar environment.

• Understanding of APIs, data handling, and backend/frontend workflows.

• Experience with LLMs, SLMs, model training and pre-training.

• Ability to convert complex ideas into structured notes, comparisons, diagrams, product requirements, and testable hypotheses.

• Strong analytical thinking, intellectual curiosity, clear written communication, and a detail-oriented approach to evidence.

• Awareness of, and commitment to, privacy, ethics, uncertainty communication, and safe use of health information.

Preferred

• Experience with LangChain, LlamaIndex, CrewAI, AutoGen, or similar agent frameworks.

• Experience building small demos using Streamlit, Gradio, Flask, FastAPI, or comparable tools.

• Basic knowledge of healthcare, chronic disease, diabetes, fitness, wearables, behavioral health, digital health, or public health.

• Understanding of data visualization, exploratory data analysis, research benchmarking, or experimental design.

• Familiarity with wearable data such as steps, sleep, heart rate, glucose, activity, symptoms, medication, or food logs.

• A portfolio (GitHub, papers, literature reviews, prototypes, or projects) demonstrating relevant work.

Expected deliverables

• Weekly summaries of relevant papers, findings, limitations, and product implications.

• Structured comparison notes across AI Hospital, Agent Hospital, PHIA, Doctor-R1 or comparable medical-reasoning systems, Personicle, and PCU.

• MVP concepts, feature breakdowns, workflows, architecture diagrams, and evaluation plans.

• Python-based demos or experiments for selected research ideas.

• Synthetic datasets or mock user profiles for safe prototype testing.

• Documentation explaining how each prototype works, what evidence supports it, its limitations, and what should be improved next.

What you'll gain

• Direct mentorship from AI researchers, engineers, clinicians, and health-domain experts.

• Exposure to end-to-end health-AI research, from literature review through experimentation and MVP translation.

• Influence and shape the design and user experience of products that could contribute to India's future health-tech stack.

• A completed research and prototype deliverable suitable for a portfolio or future publication work.

• Certificate and potential pathway to full-time roles based on performance.

Governance & data handling

This internship may involve sensitive health information. Interns will work only with de-identified or synthetic data under defined access controls and a named supervisor, following IAIRO's data-governance policy, India's DPDP framework, and applicable ICMR ethical guidance for AI in biomedical research.

Equal opportunity & fair hiring

IAIRO is an equal-opportunity employer and does not discriminate on any characteristic protected by law. We never charge fees at any stage of recruitment; any such request is fraudulent and should be reported.