# Nikhil Sai Pagidimarri — Complete Technical Context

> AI/ML Engineer & Co-founder at Kraionyx AI, specializing in deep learning, high-assurance agentic AI systems, RAG architectures, and production deployment achieving 95%+ accuracy. Published researcher in reinforcement learning.

---

## 1. Profile & Company

- **Name:** Nikhil Sai Pagidimarri
- **Role:** Co-founder, Kraionyx AI
- **Company:** Kraionyx AI (https://kraionyx.com) — High-Assurance Agentic AI Systems for Healthcare
- **Products:** KareOS (Clinical Decision Support Platform), Doclave (Doctor Collaboration Network), Svaani (Ambient Clinical Voice AI)
- **Location:** Hyderabad, India
- **Email:** nikhilsaipagidimarri@zohomail.in / nikhilsai384@gmail.com
- **Website:** https://nikhilsaipagidimarri.dev/
- **LinkedIn:** https://linkedin.com/in/nikhilsaipagidimarri
- **Twitter:** https://twitter.com/nikhiluuuuuuuuo
- **GitHub:** https://github.com/Nikhiluuuuuuuuu

---

## 2. Technical Stack

- **ML & AI:** TensorFlow, PyTorch, Scikit-learn, Keras, OpenCV, Deep Learning, Computer Vision, NLP, Reinforcement Learning, RAG, LLM Applications, Attention Mechanisms, Multi-Agent AI Systems
- **MLOps:** Docker, FastAPI, Git, CI/CD, Vector Databases, Model Monitoring, Mixed Precision Training, n8n
- **Languages & Tools:** Python, SQL, JavaScript, Pandas, NumPy, Matplotlib, HTML, CSS
- **Databases:** PostgreSQL, MongoDB, MySQL, Pinecone

---

## 3. Company Platforms & Projects

### KareOS — Clinical Decision Support Platform
- Clinical decision support workspace featuring 20+ specialist medical AI agents and HIPAA-conscious imaging analysis.

### Doclave — Professional Network for Healthcare
- Professional collaboration and knowledge-sharing platform for doctors.

### Svaani — Ambient Healthcare Voice Intelligence
- Ambient voice AI converting real-time doctor–patient conversations into structured clinical notes.

### Intelligent RAG Conversational AI Chatbot
- Built production RAG pipeline indexing 10,000+ documents with MySQL ETL.
- Enabled sub-second query responses using Pinecone and OpenAI embeddings.
- Achieved 95%+ accuracy, 40% higher relevance, and 60% fewer hallucinations.

### Brain Tumor Segmentation using ResUNet+
- Multi-class brain tumor segmentation on BraTS20 MRI dataset.
- Reached 98.17% accuracy and 0.7570 Dice coefficient using TensorFlow/Keras.
- GPU mixed precision training; inference under 2 seconds per scan.

---

## 4. Research & Publications

### Hybrid MPC and PPO Autonomous Rocket Landing (2025)
- Combined Model Predictive Control with deep RL in JAX differentiable physics.
- Reached 94.7% landing success rate with 0.62m position accuracy.
- Cut computational training time by 62% over pure RL.

---

## 5. Professional Experience & Education

### Machine Learning Intern — NextHub Technologies Pvt Ltd
- July 2024 – Oct 2024 | Hyderabad, India
- Built Smart Diabetes Management System using ensemble Combined Stacking Model.
- Achieved 92% prediction accuracy, surpassing baselines by 8%.

### Education & Certifications
- **B.Tech in AI & Data Science (2021 – 2025):** KLEF, Hyderabad | CGPA: 7.8/10.0
- **Certifications:** OCI GenAI Professional, Columbia+ ML & Prompt Engineering, Azure Fundamentals, GitHub Foundations, McKinsey Forward.