Exploring Multimodal RAG Systems
A deep dive into MRAG architectures, embedding strategies, and hybrid re-ranking pipelines for retrieval-augmented generation.
Read on MediumAvailable for opportunities
Hello, I'm
Gulshan
Singh
AI Engineer · ML Developer · Computer Vision
Building Intelligent Systems with AI, Machine Learning & Computer Vision
I am an Information Technology student passionate about building AI-powered systems that solve real-world problems. My work focuses on machine learning, computer vision, and deploying intelligent solutions.
01 / About
Who I Am
I'm Gulshan Singh, an Information Technology student at ABES Institute of Technology, passionate about building AI-powered systems that solve real-world problems. Currently serving as AI Lead at GeoPathos, a DPIIT-recognized HealthTech startup.
My expertise spans machine learning, computer vision, and generative AI — from training YOLOv11 models for real-time safety detection to researching multimodal RAG pipelines. I believe great AI engineering means bridging research and deployment.
When I'm not training models, I'm solving algorithmic problems on CodeChef (500+ problems, Gold Badge) or contributing as a GDGC Core Technical Team Member.
3+
Internships
Industry experience
3+
Projects
AI/ML applications
500+
Problems Solved
CodeChef
7.1
CGPA
ABES Institute of Technology
02 / Skills
Technical Stack
{ }
Languages
PythonC++
⚙
Machine Learning
Supervised LearningModel EvaluationFeature EngineeringPredictive Analytics
👁
Deep Learning & CV
CNNTransformers (Vision)OpenCVImage ProcessingYOLOv11
✦
Generative AI
RAGMultimodal RAGEmbeddingsMRAG Systems
⚡
Backend & Deploy
FlaskFastAPIStreamlit
🛠
Tools
DockerGitGitHubLinux
∑
Core CS
Data Structures & AlgorithmsObject-Oriented Programming
03 / Experience
Work History
AI Lead
GeoPathos Private Limited
DPIIT Recognized HealthTech Startup
July 2025 – Present
Ghaziabad, Uttar Pradesh, India
- ▸Leading AI model development for real-time healthcare monitoring solutions.
- ▸Designing and optimizing ML pipelines for predictive health analytics.
- ▸Researching integration of Computer Vision modules for advanced monitoring use-cases.
- ▸Exploring facial recognition systems for secure patient identification workflows.
- ▸Represented startup at AI Summit – Chandigarh University.
AI/ML Research Intern
Overload Ware Labs AI
June 2025
Remote
- ▸Conducted research on Multimodal Retrieval-Augmented Generation (MRAG) systems.
- ▸Analyzed retrieval strategies including shared embedding space, grounding-to-text, and hybrid re-ranking pipelines.
AI Trainee Intern
TechSaksham
Microsoft & SAP CSR Initiative
February 2025 – March 2025
Remote
- ▸Completed AI internship and worked on deep learning model development, preprocessing, and evaluation.
- ▸Gained hands-on experience in deep learning model development and evaluation.
- ▸Implemented CNN-based image classification system as part of structured AI program.
04 / Projects
Featured Work
PPE Violation Detection System
Real-time safety compliance system using YOLOv11s that detects PPE violations across 4 classes (Helmet, No Helmet, Jacket, No Jacket). Trained for 50 epochs achieving 0.89 mAP@0.5 and best F1-score of 0.89.
0.89 mAP@0.5
F1-score: 0.89
66 annotated images
Real-time inference
YOLOv11PythonOpenCVComputer Vision
Potato Leaf Disease Detection
CNN-based image classifier detecting Early Blight, Late Blight, and Healthy potato leaves. Features a Streamlit web app for real-time prediction with preprocessing pipeline.
92% validation accuracy
1500 images
3-class dataset
Streamlit UI
CNNTensorFlowStreamlitPythonOpenCV
Loan Navigator
End-to-end ML web application for loan approval prediction and CIBIL score estimation. Features secure user authentication, multi-page Streamlit interface, and real-time predictions.
Loan approval prediction
CIBIL score estimation
SQLite auth
Multi-page UI
Scikit-learnStreamlitSQLitePythonPandas
05 / Blog
Blog & Insights
Building Real-Time PPE Detection with YOLOv11
How I built a safety compliance system using YOLOv11, training on custom-annotated industrial datasets.
Read on MediumFrom CNN to Production: Potato Disease Detection
End-to-end walkthrough of training a CNN plant disease classifier and deploying it as a Streamlit app.
Read on Medium06 / Certifications
Credentials
🎓
Applied Machine Learning using Python
NIT Kurukshetra · 2025
🎓
AI for Face Recognition & Object Detection Bootcamp
Smarted Innovations · 2025
07 / Coding
Coding Profiles
👨💻
CodeChef
abit231520081
500+
Problems Solved
🥇
Gold Badge
Problem SolverGold Badge500+ Rated ProblemsConsistent Coder
08 / Contact
Get In Touch
Have a project in mind, a collaboration idea, or just want to chat about AI? Feel free to reach out.
gulshan.sg.12@gmail.com
gulshansingh125
GitHub
GulshanSingh125
Looking to hire? Download my resume directly.
Download Resume