Building intelligent systems that see, learn, and explain.

AI/ML Engineer & Builder
Lucknow, Uttar Pradesh, India
arshdeep.yadav.work@gmail.com
CSE graduate (AI/ML Honors) → incoming MBA candidate at IIT Patna (Gen AI & Product Management). Shipping across computer vision, explainable AI, robotics, and fintech.
Project Index [06] Entries
Explainable Vision Transformer for Satellite Analysis
XAI framework for satellite image classification using Vision Transformers with Grad-CAM interpretability. Published at IEEE RMKMATE'26.
2026
Real-Time IPO Tracking Platform
Full-stack fintech platform for live IPO data workflows with real-time tracking, built during Bluestock Fintech internship.
2025
Plant Disease Detection System
CNN-based crop disease detection with Streamlit interface for real-time prediction and visualization from image datasets.
2025
Deep Learning Object Detection
Real-time object detection and classification system using YOLO architecture with OpenCV tracking and transfer learning.
2024
Stress Detection via Medical Data
ML models for stress prediction using physiological datasets with feature engineering and classification optimization.
2023
IoT-Based Smart Systems
Suite of embedded systems — Smart Blind Stick, GPS Tracker, Room Monitor — built with ESP32, Arduino, and wireless protocols.
2023
← BACK TO INDEX

Explainable Vision Transformer for Satellite Analysis

Role
Lead Researcher
Timeline
2025 — 2026
Domain
Explainable AI · Remote Sensing
Tech Stack
Vision Transformers, Grad-CAM, Python, PyTorch

Problem Space

Satellite image classification is increasingly powered by deep learning, but most models operate as black boxes. Decision-makers in urban planning, agriculture, and defense need to understand why a model classifies a region a certain way — not just the label.

Traditional CNNs lack the global context needed for large satellite images, and their feature maps don't provide human-readable explanations. Vision Transformers capture long-range spatial dependencies but their attention mechanisms are opaque.

The core challenge: build a classification pipeline that achieves state-of-the-art accuracy on satellite imagery and generates interpretable visual explanations that domain experts can trust and audit.

Model Architecture

┌──────────────────────────────────┐ │ SATELLITE IMAGE INPUT │ │ (Multi-spectral Patches) │ └───────────────┬──────────────────┘ │ ┌───────────────▼──────────────────┐ │ VISION TRANSFORMER (ViT) │ │ ┌────────────────────────────┐ │ │ │ Patch Embedding (16×16) │ │ │ │ + Positional Encoding │ │ │ ├────────────────────────────┤ │ │ │ Multi-Head Self-Attention │ │ │ │ × 12 Transformer Blocks │ │ │ ├────────────────────────────┤ │ │ │ [CLS] Token → Classifier │ │ │ └────────────────────────────┘ │ └───────────────┬──────────────────┘ │ ┌───────────────▼──────────────────┐ │ EXPLAINABILITY LAYER │ │ ┌──────────┐ ┌──────────────┐ │ │ │ GRAD-CAM │ │ ATTENTION │ │ │ │ Heatmaps │ │ ROLLOUT │ │ │ └──────────┘ └──────────────┘ │ │ INTERPRETABLE OUTPUT │ └──────────────────────────────────┘

Impact

IEEE
To Be Presented
ViT
Architecture
XAI
Grad-CAM Integrated
2026
RMKMATE'26 Conf

Vision & Roadmap

Strategic Thesis

The next decade of AI impact won't come from bigger models — it'll come from intelligent integration. The gap between research breakthroughs and real-world deployment is where builders create the most value.

I focus on applied AI that crosses domain boundaries: computer vision for agriculture, explainable AI for defense and policy, embedded intelligence for accessibility, and data-driven analytics for business decisions.

My conviction: the most impactful AI systems are the ones users can trust, understand, and correct. That's why explainability, robustness, and practical deployment are central to every project I build.

Operating Principles

1. Build to deploy, not to demo. Every model should work outside a Jupyter notebook. If it can't run on a Streamlit app or an ESP32, it's not done.

2. Explain what you build. Black-box AI erodes trust. Integrate interpretability (Grad-CAM, attention visualization, SHAP) as a first-class feature, not an afterthought.

3. Cross-pollinate domains. The best solutions come from applying techniques across boundaries — computer vision for agriculture, NLP for healthcare, embedded systems for accessibility.

4. Ship fast, measure everything. Agile sprints, version control, CI/CD pipelines. Treat ML projects with the same engineering rigor as production software.

Growth Roadmap — 2025-2027 Horizons
Horizon Initiative AI Capability Focus Area Confidence
H1 — NOW Explainable AI Research Vision Transformers + Grad-CAM for interpretable classification IEEE publication; remote sensing; satellite imagery [████░]
H1 — NOW Full-Stack AI Product Development End-to-end ML pipelines with web deployment (Streamlit, Flask) AgriTech, FinTech, HealthTech applications [████░]
H2 — NEXT Generative AI Systems LLM fine-tuning, RAG pipelines, prompt engineering at scale AWS GenAI certified; applied use cases [███░░]
H2 — NEXT Hybrid MBA — Gen AI & Product Management AI product strategy, go-to-market, stakeholder coordination, data-driven roadmapping IIT Patna (Jul 2026 — 2028); McKinsey Forward; Google PM certified [█████]
H3 — LATER Autonomous Systems & Robotics Perception + planning stacks for autonomous navigation Computer vision + embedded systems convergence [██░░░]
H3 — LATER Space Technology AI ML for satellite data analysis, orbital mechanics optimization ISRO IIRS certified; remote sensing research [█░░░░]
Certifications & Credentials
Amazon Web Services
AWS Academy Graduate — Generative AI Foundations
2026
McKinsey & Company
McKinsey Forward Program
2025
Microsoft
Azure AI Fundamentals (AI-900)
2024
Google
Foundations of Project Management
2025
Microsoft
Generative AI for Data Analysis
2025
University of Michigan
Applied Machine Learning in Python
2024
IBM
Computer Vision and Image Processing
2025
ISRO IIRS
Archival & Access of Space Science Data
2025
Chandigarh University
Artificial Intelligence Analyst
2025
Chandigarh University
Predictive Analytics using IBM SPSS Modeler — Advanced
2024

Dossier

Experience Ledger
May — Jun 2025
Software Development Intern
Bluestock Fintech · Remote
Contributed to development of a real-time IPO tracking platform using Python and JavaScript
Worked on feature integration across frontend and backend systems for live financial data workflows
Collaborated with engineering and product teams during sprint-based development cycles
Supported debugging, testing, and optimization of user-facing platform components
Feb — Mar 2025
AI Data Analyst Intern
Excelerate × Rochester Institute of Technology · Remote
Cleaned and engineered datasets to improve predictive analysis workflows
Applied machine learning techniques for student engagement and churn prediction
Developed visualizations and analytical insights using Python-based libraries
Presented findings and recommendations based on trend analysis and model outputs
Jan — Feb 2025
Data Analyst Intern
Excelerate × Saint Louis University · Remote
Analyzed social media campaign datasets to identify optimization and cost-saving opportunities
Built dashboards and reports for stakeholder presentations and strategic insights
Generated actionable recommendations using data-driven analysis techniques
Dec 2024
Associate Project Management Intern
Excelerate × Saint Louis University · Remote
Developed roadmap, budgeting, and execution strategy for a global hackathon initiative
Coordinated workflows using Jira and Kanban methodologies
Evaluated project feasibility, risks, and resource allocation strategies
Presented structured project proposals to stakeholders and leadership teams
Research
2025 — 2026
Explainable Vision Transformer for Satellite Image Analysis
IEEE RMKMATE'26 International Conference
Developed an explainable AI framework for satellite image classification using Vision Transformers
Integrated Grad-CAM based interpretability methods to improve transparency in model predictions
Presented research work at IEEE RMKMATE'26 International Conference
Focused on improving trust and interpretability in AI-driven remote sensing systems
Education
Jul 2026 — 2028
Hybrid MBA — Gen AI & Product Management
Indian Institute of Technology (IIT) Patna
2-year hybrid program focused on Generative AI and AI Product Management
Incoming — starting July 2026
2022 — Jun 2026
B.E. Computer Science Engineering (Honors in AI/ML)
Chandigarh University · Mohali, Punjab
Machine Learning, Deep Learning, Computer Vision, Artificial Neural Networks
Robotics, Data Analytics, Data Structures & Algorithms
Skill Matrix
AI / Machine Learning
TensorFlow, PyTorch, Scikit-learn, OpenCV, YOLO, CNNs, NLP Fundamentals, Vision Transformers, Grad-CAM, Explainable AI
Programming
Python, SQL, Java, C/C++, JavaScript, Git, GitHub, Docker
Data & Analytics
Pandas, NumPy, Matplotlib, Tableau, Data Visualization, Predictive Analytics, Feature Engineering
Cloud & Infrastructure
AWS, Microsoft Azure, Google Cloud Platform, Streamlit, Flask, Docker, CI/CD
Product & Project Management
Jira, Agile/Scrum, Kanban, Product Research, Roadmapping, Stakeholder Coordination
Robotics & Embedded
Arduino, ESP32, IoT Systems, Sensor Integration, Wireless Communication, MQTT
┌───────────────┐ │ ┌───────┐ │ │ │ ● ● │ │ │ │ ─── │ │ │ └───────┘ │ │ ┌─┴───────┴─┐ │ │ │ HELLO! │ │ │ │ I BUILD │ │ │ │ ROBOTS. │ │ │ └───────────┘ │ │ ││ ││ │ │ └┘ └┘ │ └───────────────┘
SECRETS: ↑↑↓↓←→←→BA · TRIPLE-CLICK LOGO · TYPE "LAUNCH" · PRESS `
arshdeep@portfolio:~$