# Prayag Ravindra Adage

Pune, India | adageprayag@gmail.com | +91-8010212109
[LinkedIn](https://linkedin.com/in/prayag-adage-b7b892249) | [GitHub](https://github.com/prayagadage)

---

## Summary

Undergraduate AI & Data Science Engineer with hands-on experience leading interdisciplinary research teams to build and deploy real-world ML systems. Skilled in applying rigorous data science to scientific and research problems from medical diagnostics to cheminformatics driven drug discovery with a focus on model accuracy, reproducibility, and deployment on constrained hardware. Eager to contribute to large-scale scientific computing environments where precision and data quality drive outcomes.

---

## Education

**Bachelor of Engineering – Artificial Intelligence & Data Science** | 2022 – 2026
Marathwada Mitra Mandal's College of Engineering, Pune | Savitribai Phule Pune University
CGPA: 9.2

**Class 12 – Science** | 2020 – 2022
Sangmeshwar Junior College, Solapur
80%

---

## Experience

### Machine Learning Engineer Intern — Codeflow Studios
*6 months*

- Developed a MobileNetV3 Large computer vision model for automated crack detection and severity grading on construction walls, enabling real-time field assessment directly on embedded hardware.
- Replaced manual structural inspection workflows, reducing the client company's inspection costs by 25% and significantly cutting time-on-site per assessment.
- Managed the full ML pipeline — data collection, augmentation, training, hyperparameter tuning, and hardware deployment — integrating the model into a production system via REST API.

---

## Research & Projects

### Eye Disease Detection System – Affordable Diagnostic Tool for Rural India
*Python, PyTorch, MobileNetV3, OpenCV, Albumentations, Pillow, scikit-image*
**Research Project | Deployed**

- Led technical development in a 6-person interdisciplinary research team, managing architecture decisions, training pipeline, and code coordination via GitHub.
- Trained MobileNetV3 Large on 85,000 retinal images; improved model accuracy from 60% to 93% through iterative experimentation and targeted augmentation strategies.
- Designed for deployment on low-cost hardware to serve rural practitioners — model is live and accessible beyond the research group.

### Chemical Compound Activity Predictor – Alzheimer's Drug Discovery
*Python, RDKit, LightGBM, scikit-learn, Optuna, AutoDock Vina, pandas, NumPy*
**Research Project**

- Led technical work in a 7-person team to build a regression model predicting pIC50 values of novel compounds for Alzheimer's target inhibition, following reproducible research practices with rigorous cross-validated evaluation.
- Extracted molecular fingerprints and descriptors (Morgan, MACCS, LogP, TPSA) using RDKit; optimised LightGBM with Optuna (100 trials), achieving CV R² = 0.87 and 94.4% accuracy within ±1.0 pIC50 — up from Random Forest baseline R² = 0.77.
- Key insight: targeted removal of 5% noisy samples improved R² from 0.83 to 0.87, demonstrating that data quality outweighs model complexity — results validated through AutoDock Vina molecular docking simulations.

### Virtual Fitness Assistant – AI Posture Correction for Mobile
*Python, TensorFlow Lite, YOLO, OpenCV, MediaPipe*
**Patent Filed (2022) | Research Project**

- Conducted research under a healthcare expert team focused on posture correction; responsible for designing and deploying the full posture detection and correction software pipeline.
- Trained a YOLO-based model on TensorFlow Lite, selected for accuracy and compatibility on mobile hardware — improved pose detection accuracy from 68% to 94.45% over prior mobile-compatible baselines.
- Resolved class imbalance across 12 posture categories through targeted augmentation and loss weighting. Bridging a gap in mobile health tooling.

---

## Technical Skills

| Category | Skills |
|---|---|
| **Languages** | Python, SQL |
| **Deep Learning** | PyTorch, MobileNetV3, CNNs, BERT, LLMs |
| **ML & Modelling** | LightGBM, XGBoost, Random Forest, Regression, Classification, Clustering, Time Series |
| **Data & Analysis** | Pandas, NumPy, Scikit-learn, Optuna, Matplotlib, Seaborn, EDA, Feature Engineering |
| **Computer Vision** | OpenCV, Pillow, scikit-image, Albumentations, imgaug, MediaPipe |
| **Cheminformatics** | RDKit, AutoDock Vina, Meeko |
| **Engineering** | Docker, REST APIs, FastAPI, Flask, Firebase, Git, GitHub, OOP |

---

## Achievements & Recognition

- 1st Place – HackFusion 1.0 International Hackathon (Healthcare & Fitness Domain)
- 1st Place – PICT International Hackathon, Pune (Healthcare Domain)
- 2nd Place – State Level Hackathon, MIT College of Railway Engineering, Barshi
- District & National Level Hackathon wins for posture correction device (5 total wins across all levels)
- Patent Filed – AI-based fitness posture correction technology (2022)

---

## Leadership & Community

- Tutored 10+ students in AI and machine learning fundamentals; Core Member of AI & Data Science Club – organised workshops and coding sessions.
- Hackathon Team Leader – represented teams across national and international competitions, achieving multiple podium finishes.
- Research Team Lead – led technical execution across two active research collaborations (eye disease detection, Alzheimer's drug discovery), coordinating cross-disciplinary teams of 6–7 members.