Deepti Bahel
Engineer of Data. Advocate for Health. Believer in Better Systems.
Actions
Deepti Bahel is a seasoned Senior Business Intelligence Engineer with over 12 years of experience transforming complex data into meaningful, actionable insights. With a Master’s in Business Analytics from Purdue University and a strong background in BI tools like Power BI, Tableau, and Looker, Deepti has led analytics transformations at companies like Intuit, Google, Wayfair, and Hanna Andersson.
As a kidney transplant recipient and health coach, Deepti brings a unique perspective to healthcare analytics—combining deep technical knowledge with a mission-driven approach to improving patient outcomes through data.
She specializes in building self-service BI platforms, streamlining data pipelines using tools like Superglue, Databricks, and Snowflake, and bridging the gap between technical teams and non-technical decision-makers. Her recent work focuses on embedding analytics into operational workflows through interactive apps powered by Streamlit and AI-driven assistants.
Deepti is also an active mentor, speaker, and advocate for data equity in healthcare, featured by media outlets like India Currents and KnowFix News, and regularly invited to speak at data and health tech panels.
Links
Area of Expertise
Building a Startup-Ready AI Product Without a Big Team: Lessons from the Frontlines of Healthcare an
People often say, “Just build an AI product.”
It sounds simple, almost casual — as if all you need is a weekend, a laptop, and a model.
But building one—truly building one—stretches every part of you. Your architecture choices get tested. Your patience gets tested. Your discipline, your stamina, your optimism… sometimes even your health.
In the last year, I built three end-to-end healthcare AI tools — NephroCompass, Carelytics (Hospital BI), and a Blood Donation Decision Support System — mostly by myself. I wrote code between doctor’s appointments. I trained models while waiting for lab results. I designed product flows on days when recovery felt heavier than usual. And I did it while job searching, rebuilding my life after a kidney transplant, and trying to create something that might make the journey a little easier for the next patient.
What this year taught me is that you don’t need a large team or a perfect dataset to build something meaningful. What you do need is clarity, courage, and the humility to let constraints become your best teammates.
This talk isn’t a highlight reel. It’s the honest, backstage story of what it takes to build startup-ready AI products when it’s just you, your determination, and a problem you refuse to ignore.
• How to design an MVP that doesn’t fall apart the moment a clinician or patient touches it
• How to move fast when your data is limited, messy, emotional, and deeply human
• How to label data without a room full of annotators
• How to evaluate models when “good enough” isn’t good enough — not in healthcare
• How to choose deployments that make sense when you’re one person, not a 50-engineer team
• How to define ROI early so your product doesn’t drift into “cool demo” territory
• And the part no one talks about: building while healing, while doubting, while starting over — and still believing that it matters
This isn’t a product demo.
It’s a field guide — part architecture, part scrappy tactics, part survival manual — for anyone building AI in the real world with limited resources, imperfect data, and a heart full of stubborn purpose.
If attendees walk away with new frameworks, great.
But I hope they walk away with something deeper: the reminder that meaningful AI doesn’t begin with a big team or a giant budget. Sometimes it begins with one person, one lived experience, and one quiet decision to try anyway.
Agentic AI in Healthcare: Architecting Systems That Don’t Hallucinate When Lives Are on the Line
Agentic AI is everywhere right now — writing code, answering questions, routing tasks, and stitching together workflows like an eager new intern. But in healthcare, an overconfident agent that “just makes something up” isn’t being creative. It’s being dangerous.
Over the past year, I’ve built a set of healthcare agents and decision-support tools — NephroCompass, Carelytics (Hospital BI), and a Blood Donation DSS — all of which interact with sensitive data and real patient journeys. When you’re dealing with someone’s health, the room for error shrinks fast. Every hallucination isn’t just a bug; it’s a breach of trust.
This session is the honest, behind-the-scenes look at what it actually takes to make agentic AI behave like a responsible teammate instead of a chaotic one. It’s the messy middle — designing systems that stay grounded, reason in context, and hold up under pressure, even when the data is imperfect, emotional, or fragmented.
We’ll talk about:
• Why plain RAG falls short in healthcare, and what actually makes retrieval reliable
• How domain grounding, layered retrieval, and context shaping reduce hallucinations
• Architectures that keep agents safe without turning them into slow bureaucrats
• How to design step-by-step reasoning patterns that clinicians can follow and trust
• Real moments from NephroCompass and Carelytics — where the agent almost went off-track, and how we pulled it back
• How to test, validate, and monitor an agent when your “users” are patients, pharmacists, and care teams who can’t afford mistakes
• Why sometimes the right fix isn’t clever prompting — it’s solid, boring, beautiful data engineering
This isn’t a talk about futuristic magic. It’s about building AI that behaves well in the real world, especially where the stakes are personal and human.
Building agentic AI is fun.
Building agentic AI that people can trust — that’s the real craft.
This session introduces a no-code, AI-enhanced Hospital Strategy Optimization Dashboard that helps a
This session introduces a no-code, AI-enhanced Hospital Strategy Optimization Dashboard that helps administrators make cost-efficient care decisions. Built with Streamlit, the app integrates linear programming, anomaly detection (Isolation Forest), and GPT-driven strategic recommendations to evaluate three patient selection strategies: Linear Programming, Greedy, and Heuristic. Users can set real-time constraints, compare cost savings, analyze patient trends, and explore actionable strategy insights. A must-see for anyone interested in operational healthcare analytics and data-driven decision support.
Chat-Based Anomaly Detection for Hospitals: An AI-Powered Streamlit App for Insightful Audits
Discover how to detect hospital billing anomalies with a no-code, AI-powered Streamlit app. This session walks through a real-world solution combining ML models, SHAP explainability, and GPT-generated insights. Learn to automate audits, visualize outliers, and communicate findings with business users—all in one interactive dashboard. Ideal for healthcare BI teams, analysts, and data scientists.
Optimizing Healthcare Efficiency with Power BI: A Live Case Study Using Real-World Hospital Data
Optimizing Healthcare Efficiency with Power BI: A Live Case Study Using Real-World Hospital Data
Healthcare systems today are drowning in fragmented data. Billing inefficiencies, reporting delays, and manual processes cost organizations both time and money. In this talk, I’ll present a real-world solution: an end-to-end Power BI dashboard that transforms raw hospital data into actionable insights.
Built on a cleaned and enriched version of the Kaggle Hospital Dataset (2019–2024), this dashboard identifies billing anomalies, visualizes cost drivers by medical condition, and segments patient stays to uncover operational inefficiencies.
🔍 Key Features Showcased:
MoM (Month-over-Month) Billing Trend Analysis – Detect unusual cost spikes in real-time
Stay Category Segmentation – Short, Medium, Long, Very Long stays for targeted care strategy
ICD Code Mapping – Linking diagnoses to billing impact for smarter planning
Multi-Hospital Comparison – KPI breakdown across six leading institutions
Self-Service Filters – Designed for non-technical users to explore the data dynamically
⚙️ Technical Stack: Power BI, DAX, Python (data cleanup), Streamlit (embedding)
🤝 Collaborative Credit: Special thanks to Andrew Malinow for the ICD-10 integration guidance
🔊 Audience Takeaways:
How to go from raw, unstructured hospital data to a business-ready reporting tool
Strategies to visualize and act on billing inefficiencies using Power BI
The impact of accessible analytics tools in operational healthcare decision-making
Practical steps to integrate anomaly detection and forecasting into healthcare BI workflows
Deepti Bahel
Engineer of Data. Advocate for Health. Believer in Better Systems.
Links
Actions
Please note that Sessionize is not responsible for the accuracy or validity of the data provided by speakers. If you suspect this profile to be fake or spam, please let us know.
Jump to top