About Dr. Bondugula
Dr. Chandra Sekhar Bondugula is one of India's foremost practitioners at the intersection of Artificial Intelligence and medical education. As Founder and CEO of EdMedAI, he has built the country's most comprehensive AI-powered platform for Competency Based Medical Education (CBME) — aligned to the National Medical Commission's 2,683-competency framework and used by medical colleges across India.
His expertise spans the full stack of modern AI: from designing and fine-tuning Large Language Model (LLM) pipelines, building Retrieval-Augmented Generation (RAG) architectures, and engineering AI fallback systems, to applying these technologies directly to the real-world challenge of transforming Indian medical education. He is not a theorist who studies AI in medical education — he is the architect who built it.
Dr. Bondugula brings a rare dual perspective: deep roots in US graduate medical education (ACGME milestones, EPA frameworks, simulation-based training) and an intimate, ground-level understanding of the Indian CBME ecosystem — the NMC regulations, the challenges medical colleges face in implementation, and the specific AI capabilities that make compliance achievable at scale.
"AI in medical education is not about replacing the teacher. It is about giving every teacher — regardless of which tier of college they work in — the same quality of content, analytics, and student support that was previously available only at the most resourced institutions."
— Dr. Chandra Sekhar Bondugula, Founder & CEO, EdMedAIAI Model Development & Large Language Models
Dr. Bondugula's AI work is engineering-first. He designed and built EdMedAI's entire AI infrastructure — including its multi-model LLM orchestration, RAG pipeline, and custom medical AI systems — from the ground up. His hands-on LLM expertise includes:
Multi-Model LLM Orchestration
Built a 3-tier LLM fallback system: Google Gemini 2.0/2.5 Flash (primary) → OpenAI GPT-4o (fallback) → Anthropic Claude Sonnet (complex tasks). Zero downtime on AI generation across all user interactions.
Retrieval-Augmented Generation (RAG)
Designed a domain-specific RAG pipeline that uses NMC competency codes — not document chunks — as retrieval keys. This grounds every AI-generated clinical case, quiz, and lecture plan in the exact NMC competency being addressed.
AI Question Bank at Scale
Built an autonomous background AI system that pre-generates and maintains 10 MCQs per difficulty level per competency — covering all 2,683 NMC competencies = 80,490+ questions — using GPT-4o and Gemini with deduplication via MD5 indexing.
AI Prompt Engineering
Engineered domain-specific prompts for medical content generation — case studies, lecture plans, OSCE scenarios, SGD cases, ethics reflections — calibrated to NMC competency domain levels (K/KH/SH/P) and difficulty targets.
AI Feedback Triage System
Designed an LLM-based triage system with a 26-entry behavioural specification to classify user feedback as workflow confusion vs real bugs, generating corrective guidance — a novel application of LLMs in EdTech platform management.
AI Conversational Tutor
Built a persistent, competency-aware AI Tutor that maintains full conversation history per student per competency in PostgreSQL — enabling contextual follow-up questions and learning trajectory analysis across sessions.
Video Lecture AI Analysis
Developed a 3-tier RAG system for video lecture analysis — generating AI study notes, key highlights, and interactive MCQs from uploaded lecture videos using transcript extraction and LLM processing.
AI Fraud Detection Models
Designed a 6-signal unified fraud scoring model for clinical logbooks: UHID deduplication, file hash comparison, GPS hospital geofencing, EXIF timestamp validation, cross-student similarity, and AI-content detection.
Competency Based Medical Education (CBME) Expertise
Dr. Bondugula has spent years building the most comprehensive digital implementation of the NMC CBME framework in India. His CBME expertise is not academic — it is operational: he built the systems that make CBME compliance trackable, auditable, and achievable for real medical colleges with real faculty constraints.
- NMC 2,683-competency database: Built and maintains the complete NMC CBME competency database — all subjects, all phases, all domain levels (K/KH/SH/P) — as the core data model of EdMedAI
- Digital DOAP tracker: First digital implementation of the NMC-mandated DOAP (Demonstrate → Observe → Assist → Perform) pedagogy with multi-repetition Perform counting and mandatory HOD approval gate
- SM-2 Spaced Repetition for CBME: Adapted the SM-2 spaced repetition algorithm specifically for NMC competency mastery — with ease factor bounded by competency domain level and intervals gated on logbook evidence for SH/P competencies
- CBME knowledge base: Authored a 4,335-line CBME knowledge base that powers EdMedAI's RAG-based CBME Assistant Chatbot — the only platform-meta + curriculum chatbot in Indian medical education
- NMC Annexure 5 compliance: Implemented NMC 2024 CBME Annexure 5 hour targets as first-class database entities, cross-validated against actual logged teaching sessions
- AETCOM digital module: Built the digital AETCOM (Attitudes, Ethics and Communication) module — tracking attitude and ethics competencies alongside clinical competencies in the same platform
Graduate Medical Education — USA & India
Dr. Bondugula's graduate medical education expertise spans two of the world's most rigorous regulatory frameworks: the ACGME (Accreditation Council for Graduate Medical Education) in the USA, and the NMC's PGMER 2023 framework in India. This dual expertise is rare and directly informs EdMedAI's PG programme features.
USA Medical Education Experience
Dr. Bondugula's US medical education background includes working with ACGME milestone-based residency frameworks, EPA (Entrustable Professional Activities) supervision level systems, simulation-based clinical training infrastructure, and digital portfolio assessment systems. The ACGME's milestone framework — which tracks resident competency progression across six domains from sub-competency to independent practice — is the conceptual foundation for EdMedAI's competency milestone analytics and DOAP progression tracking.
His US experience also shaped EdMedAI's design philosophy: in US academic medical centres, data-driven teaching, AI-assisted content creation, and formative assessment with documented feedback are standard practice — not aspirational goals. EdMedAI was built to bring this same infrastructure quality to Indian medical colleges.
India PGMER 2023 Expertise
Dr. Bondugula built EdMedAI's complete PG Residency module aligned to NMC PGMER 2023, including: Workplace-Based Assessments (DOPS, Mini-CEX, Direct Teaching), thesis milestone tracking with document evidence chains, 7-domain quarterly formative scoring per Annexure III, guide/co-guide role separation, and PG competency framework with syllabus integration.
AI-Driven Curriculum Design
Dr. Bondugula pioneered the application of AI to medical curriculum design in India — not as a chatbot overlay on existing curriculum, but as a fundamental redesign of how curriculum content is generated, maintained, and delivered. His key innovations in AI curriculum design include:
Competency-Grounded Content Generation
AI content generation (case studies, lecture plans, MCQs, SGD cases) anchored to live NMC competency codes — not generic medical topics. Every AI-generated piece is traceable to a specific competency and domain level.
Adaptive Learning Pathways
Spaced repetition quiz scheduling adapted to competency domain — easier and faster progression for K-domain knowledge, logbook-gated intervals for SH/P-domain skills. AI personalises the learning path per student.
AI Clinical Simulation
Built 50+ interactive clinical simulators — ECG interpretation, anatomy 3D, pathology slides, virtual patient encounters — that provide simulation-based learning before patient contact, bridging the pre-clinical to clinical gap.
Analytics-Driven Curriculum Improvement
Learning analytics at the competency level — which competencies students consistently fail, which teaching methods produce the best outcomes — feeding back into curriculum delivery recommendations for faculty and HODs.
11 Provisional Patent Innovations
Following a comprehensive codebase review in April 2026 covering 3,023 schema lines, 32,123 routes lines, 9,602 storage lines, and 121 frontend pages, Dr. Bondugula identified 11 distinct provisional patent innovations embedded in EdMedAI's architecture:
| # | Innovation | Core AI/Technical Hook |
|---|---|---|
| 01 | Ghost Faculty Prevention | GPS velocity physics + MCR global uniqueness + device fingerprinting + Aadhaar hash + impossible-velocity detection |
| 02 | Multi-Layer Clinical Logbook Fraud Detection | 6-signal unified AI fraud score: UHID dedup + file hash + GPS geofence + EXIF timestamp + cross-student similarity + AI-content detection |
| 03 | Context-Aware AI Feedback Triage | 26-entry behavioural spec injected into LLM; distinguishes workflow confusion from real bugs; generates corrective guidance automatically |
| 04 | RAG-Enhanced NMC Competency-Grounded Content Generation | Live NMC competency code as RAG retrieval key (not document chunks); generation differentiated by K/KH/SH/P domain level |
| 05 | Digital DOAP Tracker with Multi-Rep Perform Counting | First digital NMC DOAP implementation; multi-rep Perform counting; mandatory HOD approval gate before certification |
| 06 | SM-2 Spaced Repetition Adapted for NMC Competency Mastery | Ease factor bounded by competency domain level; SH/P intervals gated on logbook evidence; adaptive scheduling per student |
| 07 | Ephemeral Session-Code Attendance | Academic-year eligibility enforcement before attendance accepted; dual geofence modes (radius + polygon); NMC 75%/80% targets built into model |
| 08 | 5-Tier Medical Education Regulatory Hierarchy | NMC 2024 CBME Annexure 5 hour targets as first-class queryable DB entities; cross-validated against actual logged sessions |
| 09 | Clinical Encounter Simulator with Branching Scenarios | DB-editable JSONB branching decision trees; scoring across 6 NMC clinical reasoning domains; mapped to competency codes |
| 10 | CBME Platform Chatbot with 4,335-line Knowledge Base | Platform-meta + curriculum chatbot; conversation persistence tied to competency context; full-history escalation pathway |
| 11 | PG Residency Milestone Tracking with Thesis Evidence Chain | Formal review-request workflow per milestone with document evidence; 7-domain quarterly formative scoring per Annexure III |
Research & Thought Leadership
Dr. Bondugula is an active contributor to the discourse on AI in medical education — through EdMedAI's published knowledge base, through speaking on the intersection of LLMs and medical curriculum, and through the practical research embedded in every feature of EdMedAI's platform. His thought leadership spans three intersecting domains:
- AI safety in medical education: How to use LLMs responsibly — with competency grounding, hallucination mitigation through RAG, and human-in-the-loop validation — so that AI-generated medical content is clinically accurate
- LLM capabilities and limitations in healthcare: Which tasks LLMs do well in medical education (content generation, question creation, feedback classification) and which require human expertise (clinical sign-off, ethical judgment, formative feedback quality)
- Bridging US and Indian medical education: How ACGME milestones, EPA frameworks, and simulation-based training can be adapted to the Indian CBME regulatory context — taking the best of both systems
Every insight Dr. Bondugula publishes has been tested in production — in a live platform used by real students and faculty in real Indian medical colleges. His AI work is not research-paper AI; it is deployed, monitored, and continuously improved AI that must work every day for thousands of users.
Expert Content by Dr. Bondugula
The following articles are researched and written by Dr. Bondugula, drawing on his experience building EdMedAI and his expertise in AI, LLMs, CBME, and graduate medical education: