Building intelligentsystems thatmatter
AI/ML Engineer developing scalable AI/ML and intelligent systems that redefine how organizations process and understand information.
About
I focus on advancing memory-augmented LLMs and AI agents, currently building project-specific hierarchical memory systems for more coherent, long-term reasoning.
My past work spans vision models, multilingual RAG systems, and large-scale regressors, shaping my interest in scalable intelligence and how models store, update, and use structured knowledge effectively.
Featured Projects
Hybrid ViT Model for Crop Diagnosis
Nov 2025Advanced Crop Disease Detection System
Designed a hybrid ViT + CNN + Custom MLP architecture that processes images in parallel to detect 38 crop disease classes, achieving 85% training accuracy and 99.5% validation accuracy. Developed an IoT-ready inference pipeline for real-time deployment with sensor integration and filed a patent covering the model architecture and end-to-end diagnostic workflow.
Key Metrics
ORIN Tri-Sense AI
Oct 2025Multilingual RAG Chat-Runtime
Built and fine-tuned a multilingual RAG chat-runtime for universities, government, and private orgs (6+ languages). Optimized proprietary RAG (Pinecone) with tri-source retrieval and fine-tuning, reaching 86% relevance and 0.72 MRR. Scaled to 1k sessions / 200 QPS (latency<300 ms), reducing time-to-answer from 96 hours to 10 minutes.
Key Metrics
Product Price Regressor (DeBERTa-v3)
Oct 2025Price Prediction System
Designed and iteratively fine-tuned a DeBERTa-v3-base regression model over 30 epochs (15+7+8) on a 75,000-sample dataset for price prediction, ultimately achieving a final best validation metric of SMAPE 21.73% (down from initial 63.17%). Optimized training efficiency and stability on a Tesla P100-PCIE-16GB GPU using Automatic Mixed Precision (FP32-FP16), a 2e-5 learning rate, 0.01 weight decay, and gradient clipping (max_norm=1.0), with final architecture size being 183 Million Parameters.
Key Metrics
Skills & Technologies
Languages
AI/ML Frameworks
Platforms & Tools
Education & Certifications
Education
Bachelor of Technology
Computer Science and Engineering
Lovely Professional University • CGPA 8.33 • 2022-2026
Minor: Machine Learning and Deep Learning
Senior Secondary Education
Physics, Chemistry, Math, Computer Science
Sainik School Kapurthala • 2015-2022
Certifications & Achievements
Building Intelligent Troubleshooting Agents
Microsoft • Sep 2025
AI and Machine Learning Algorithms
Microsoft • Sep 2025
Generative AI with Large Language Models
DeepLearning.AI • Apr 2024
Microsoft Founder Hub Member
$20,000 in benefits and cloud credits
NCC A & B Certificates
6 years of service