Machine Learning Engineer | Published Researcher (IEEE CICN-2025) | M.Tech in Applied AI @ VNIT Nagpur

Specializing in production ML systems, robust deep learning, and healthcare AI. I build end-to-end ML pipelines from research to deployment, with expertise in time-series modeling, computer vision, and failure-aware systems.

• 🔬 ML Research → Published at IEEE CICN-2025, journal submission under review
• 🏗️ Production ML → Building robust, deployable ML systems with MLOps best practices
• 🏥 Healthcare AI → Computer vision for medical imaging with explainable AI
• 📊 Data Engineering → End-to-end pipelines with Kafka, Airflow, and AWS
• 🚀 Deployment → Containerized ML services using BentoML, Docker, Kubernetes

Robust RSSI-based localization achieving 1.3% degradation under 92% sensor failure (vs 2614% in baselines)

A Sparse Signal ML framework using Masked Channel Pretraining and Attention Mechanisms to enable reliable indoor localization when infrastructure fails. Solves a critical real-world problem where standard deep learning models catastrophically collapse.

🔑 Key Innovations:

• Masked Channel Pretraining (MCP) - Self-supervised learning for robust representations
• Conditional WGAN-GP - Constrained augmentation for rare failure patterns
• TCN-Transformer Hybrid - Local stability + global attention for partial observability
• Production-Ready - Sub-50ms inference on commodity hardware

📊 Impact: 20× improvement over industry baseline (1.3% vs 20-40% degradation under failures)

🛠️ Tech Stack: PyTorch, Transformers, WGAN-GP, TCN, Attention Mechanisms, MLflow

87.7% classification accuracy for sub-meter indoor positioning using BLE RSSI fingerprints

End-to-end MLOps project demonstrating research-to-production pipeline for indoor localization using deep learning.

🔑 Key Features:

• Hybrid Architecture - CNN spatial features + LSTM temporal modeling
• Ensemble Strategy - 5-fold cross-validation with logit averaging
• Production Deployment - BentoML service with REST API
• Containerized - Docker + Kubernetes ready
• Experiment Tracking - Full MLflow integration

🛠️ Tech Stack: PyTorch, BentoML, Docker, MLflow, Kubernetes

📄 Publication: IEEE Xplore
📚 Citation: Vansarla & Agarwal, IEEE CICN 2025, DOI: 10.1109/CICN67655.2025.11367998

Vision-Guided Clinical Summary System combining computer vision + NLP for automated medical documentation

Healthcare AI application integrating fundus image classification with GradCAM explainability and Retrieval-Augmented Generation (RAG) for patient-friendly clinical summaries.

🔑 Key Features:

• Explainable AI - GradCAM visual explanations for medical image classification
• RAG Pipeline - Context-aware clinical summary generation
• Modular Architecture - Clear API interfaces for production deployment
• Safety-First Design - Safe prompt engineering for non-diagnostic clinical support

🛠️ Tech Stack: PyTorch, OpenCV, Transformers, RAG, FastAPI, GradCAM

🎯 Use Case: Automated clinical documentation reducing physician workload while maintaining explainability

End-to-end NLP pipeline from web scraping to sentiment visualization

Complete data pipeline demonstrating: web scraping (BeautifulSoup) → text processing → sentiment analysis (TextBlob) → data visualization.

🛠️ Tech Stack: Python, BeautifulSoup, TextBlob, Pandas, Matplotlib

PyTorch TensorFlow Scikit-learn Transformers Hugging Face LangChain

Specialized: CNN-LSTM, Attention Mechanisms, GANs, Masked Learning, Self-Supervised Learning, RAG

OpenCV YOLO GradCAM BERT GPT Explainable AI

MLflow BentoML Docker Kubernetes CI/CD Model Deployment API Development

Apache Kafka Apache Airflow Terraform AWS Glue AWS Kinesis MySQL PostgreSQL NoSQL

Certified: DeepLearning.AI Data Engineering Specialization (Dec 2025)

AWS (S3, EC2, RDS, Glue, Kinesis) REST APIs DAG Orchestration ETL Pipelines

Python SQL C++ Git NumPy Pandas Matplotlib Seaborn

📄 Anil Vansarla, Amit Agarwal
"AI-Based Indoor Localization Using RSSI: CNN–LSTM Based Approach"
IEEE 17th International Conference on Computational Intelligence and Communication Networks (CICN-2025)
Goa, India | December 20-21, 2025
DOI: 10.1109/CICN67655.2025.11367998

📄 Anil Kumar, Amit Agarwal
"Failure-Aware RSSI-Based Indoor Localization Using Masked Attention Models"
Submitted to SCI-Indexed Journal | January 2026
Achieving 1.3% degradation under 92% sensor failure vs 2614% in supervised baselines

M.Tech in Applied Artificial Intelligence
Visvesvaraya National Institute of Technology (VNIT), Nagpur | 2024-2026 | CGPA: 8.37/10.0

B.E. in Electrical & Electronics Engineering
Osmania University, Hyderabad | 2012-2016

• ✅ Data Engineering Specialization - DeepLearning.AI & AWS (Dec 2025)
• ✅ Data Modeling, Transformation, and Serving - DeepLearning.AI (Dec 2025)
• ✅ Data Storage and Queries - DeepLearning.AI & AWS (Nov 2025)
• ✅ Source Systems, Data Ingestion, and Pipelines - DeepLearning.AI & AWS (Oct 2025)

Technical Content Developer (AI/ML Intern)
NxtWave Disruptive Technologies | March 2025 - September 2025

• Developed ML/DL modules for Data Science program serving 5,000+ learners
• Collaborated with industry engineers in Agile sprints, delivering 12+ technical modules
• Created Python-based demonstrations and interactive notebooks for applied ML concepts
• 95%+ learner satisfaction across 3 Agile sprints

• IEEE Published ML system deployed with BentoML
• Failure-aware localization framework under journal review
• Healthcare AI with explainable vision models
• End-to-end ML pipelines with MLOps and cloud integration

🔍 Seeking Full-Time Machine Learning Engineer Roles

Areas of Interest:

• 🏥 Healthcare AI & Medical Imaging
• 📊 Time-Series Forecasting & Sensor Data
• 🔧 Robust ML Systems & Failure-Aware Learning
• 🚀 Production ML & MLOps
• 📡 IoT & RSSI-Based Localization

📧 Email: anil.kumar87654321@gmail.com
🔗 Portfolio: https://github.com/timbersaw-jugg
📄 Resume: Available upon request

Last Updated: January 2026
© 2026 Anil Vansarla | VNIT Nagpur