Recent B.Tech graduate in Artificial Intelligence & Data Science, focused on building practical, end-to-end intelligent systems.
Hands-on experience across Machine Learning, Deep Learning, Generative AI, LLMs, RAG, AI Agents, NLP, Computer Vision, and Multimodal AI.
Experienced in developing AI applications from data preprocessing and model development to API integration, testing, deployment, and user-facing interfaces.
Built projects involving RAG-based learning assistants, multi-agent data science systems, medical AI, conversational AI, image captioning, multimodal reasoning, and full-stack AI applications.
Comfortable working with Python, SQL, TensorFlow, PyTorch, Scikit-learn, LangChain, LangGraph, FastAPI, Flask, Streamlit, React, PostgreSQL, ChromaDB, AWS, and GCP.
Published research on ECM²RS — Explainable Causal Multi-Modal Reasoning System, focusing on explainable and causal multimodal reasoning.
Interested in building reliable, explainable, scalable, and production-oriented AI systems that solve real-world problems.
Open to opportunities in AI/ML Engineering, Data Science, Generative AI, Machine Learning, and related engineering roles.
- Programming & Data: Python, SQL, R, EDA, Data Preprocessing, Feature Engineering, Data Visualization
- Machine Learning: Supervised & Unsupervised Learning, Classification, Regression, Clustering, Ensemble Learning, Model Evaluation
- Deep Learning: CNN, RNN, LSTM, Transfer Learning, Image Classification, Image Captioning
- NLP: Text Classification, Sentiment Analysis, TF-IDF, BERT, Multi-Label Classification
- Generative AI: LLMs, Prompt Engineering, RAG, AI Agents, LangChain, LangGraph, Structured Outputs
- Computer Vision: Image Processing, VQA, Multimodal Learning, LLaVA, Grad-CAM, Explainable AI
- AI Engineering: FastAPI, Flask, Streamlit, REST APIs, React, PostgreSQL, ChromaDB
- MLOps & Cloud: AWS EC2, AWS Bedrock, GCP, MLflow, Git, Model Deployment, Testing
Published in: International Journal for Research in Applied Science & Engineering Technology (IJRASET), 2026
ECM²RS is an explainable multimodal reasoning framework designed to address the black-box behavior of AI models by integrating visual understanding, language reasoning, causal analysis, and explainability techniques to make multimodal decision-making more transparent and interpretable.
Research Focus :
- Addressing black-box behavior in multimodal AI systems
- Explainable AI and model interpretability
- Causal reasoning for transparent decision-making
- Multimodal visual and textual reasoning
- Visual Question Answering (VQA)
- Mitigating language and dataset biases
Core Technologies :
ResNet50 BERT LLaVA Grad-CAM Causal Reasoning
Datasets :
VQA COCO 2017 CLEVR ScienceQA
Key Contributions :
- Developed a multimodal reasoning pipeline integrating vision and language models.
- Incorporated Grad-CAM and textual explanations to improve model interpretability.
- Introduced causal reasoning to provide more transparent reasoning pathways.
- Investigated approaches for reducing the black-box nature of multimodal AI decision-making.
- Evaluated the framework across multiple multimodal reasoning benchmarks.
Publication :
- Journal: IJRASET
- Volume: 14
- Issue: IV
- Year: 2026
- DOI: 10.22214/ijraset.2026.80540
AI-powered medical intelligence platform for chest X-ray analysis, disease prediction, explainable AI, and AI-assisted medical reporting.
Tech: Python EfficientNet Grad-CAM Flask REST API Groq PostgreSQL
- Trained an EfficientNet-based pneumonia classification model on chest X-ray images.
- Implemented Grad-CAM to provide visual explanations for model predictions.
- Developed REST APIs for image analysis, prediction, and explainability workflows.
- Integrated LLM-based medical report generation using structured model outputs.
- Built an end-to-end workflow connecting AI inference, explainability, APIs, database, and application interface.
RAG-based personal learning assistant that enables users to interact with their documents through semantic retrieval and context-aware AI responses.
Tech: Python FastAPI Streamlit ChromaDB Sentence Transformers RAG LLM Pytest
- Built a document ingestion and processing pipeline for learning materials and PDFs.
- Implemented text chunking, embeddings, vector storage, and semantic retrieval using ChromaDB.
- Developed context-aware question answering with source citations and conversation memory.
- Implemented relevance-based retrieval using configurable similarity thresholds.
- Added automated tests covering chunking, embeddings, retrieval, vector storage, and RAG workflows.
- Designed the system with modular services for retrieval, vector storage, and LLM interaction.
Multi-agent AI system designed to automate key stages of a data science workflow from dataset profiling to analytical insights and machine learning recommendations.
Tech: Python Generative AI AI Agents GCP Pandas Machine Learning
- Developed specialized agents for data profiling, analytics, ML recommendations, insights, and report generation.
- Automated exploratory analysis and identification of important dataset patterns.
- Integrated LLM-powered reasoning for natural-language analytical insights.
- Designed an agentic workflow for transforming raw datasets into structured analytical reports.
- Deployed the application using Google Cloud Platform.
Full-stack expense claims management platform for submitting, tracking, validating, and auditing employee expense claims.
Tech: FastAPI React PostgreSQL JWT REST API JavaScript
- Developed a FastAPI backend with authentication, claims, receipts, and audit-log workflows.
- Built a React frontend for user authentication and expense claim management.
- Implemented JWT-based authorization and protected API endpoints.
- Designed PostgreSQL data models for users, claims, receipts, and audit records.
- Connected the frontend and backend through REST APIs.
- Deployed the backend and frontend as separate web services.
AI-powered career guidance application providing personalized career recommendations, skill analysis, learning roadmaps, and actionable development plans.
Tech: Python Streamlit Gemini Prompt Engineering AWS EC2
- Built a conversational AI application with multi-turn conversation memory.
- Designed structured prompts and JSON-based outputs for consistent responses.
- Implemented guardrail classification and response validation.
- Added retry mechanisms, logging, and token usage tracking.
- Deployed the application on AWS EC2.
Research project focused on addressing the black-box behavior of multimodal AI systems through explainable and causal reasoning.
Tech: ResNet50 BERT LLaVA Grad-CAM Causal Reasoning
- Developed a multimodal reasoning framework combining visual and language representations.
- Integrated Grad-CAM for visual explanations of model predictions.
- Incorporated textual explanations and causal reasoning to improve reasoning transparency.
- Investigated multimodal reasoning across VQA COCO 2017, CLEVR, and ScienceQA datasets.
- Research published in IJRASET, 2026.
Deep learning-based image captioning system that generates natural-language descriptions from images.
Tech: Python TensorFlow VGG16 LSTM Flickr8k Streamlit
- Built an encoder-decoder image captioning architecture using VGG16 and LSTM.
- Used transfer learning for visual feature extraction.
- Trained the model on the Flickr8k dataset containing 8,000 images with multiple captions per image.
- Evaluated generated captions using BLEU score.
- Achieved a BLEU score of approximately 0.29.
- Developed a Streamlit interface for interactive image caption generation.
Generative AI application that transforms YouTube video transcripts into structured articles and downloadable websites.
Tech: Python LangChain Groq Streamlit LLMs
- Extracted and processed YouTube video transcripts.
- Used LLM-powered generation to create structured articles from video content.
- Added controls for tone, length, and generation configuration.
- Generated complete HTML/CSS/JavaScript websites from the processed content.
- Provided interactive previews and downloadable website packages.
| Project | Focus | Technologies |
|---|---|---|
| TelcoVision | Telecom Churn Prediction | XGBoost · LightGBM · ML · Streamlit |
| Flipkart Sentiment Analysis | Sentiment Classification | NLP · TF-IDF · ML · Streamlit |
| Multi-Label Toxic Comment Detection | Multi-Label NLP Classification | TF-IDF · Logistic Regression · NLP |
Innomatics Research Labs · Internship
Nov 2025 – Apr 2026
- Worked on end-to-end Machine Learning, Generative AI, RAG, and AI Agent workflows.
- Developed ML pipelines involving data preprocessing, EDA, feature engineering, model training, and evaluation.
- Built LLM-powered applications using LangChain, LangGraph, prompt engineering, and RAG.
- Worked with AI agent workflows for task automation and intelligent decision-making.
- Applied MLOps practices including experiment tracking and model deployment.
- Deployed AI/ML applications using AWS EC2 and explored cloud-based AI services.
Excelerate · Internship
Oct 2025 – Nov 2025
- Coordinated project activities involving planning, task management, resource allocation, and delivery tracking.
- Worked with cross-functional teams to organize project workflows and milestones.
- Researched and evaluated AI-powered tools for improving project management and workflow automation.
- Contributed to project documentation, risk identification, and process improvement.
Evoastra Ventures Pvt. Ltd. · Internship
Sep 2025 – Oct 2025
- Worked on data analysis and machine learning workflows for real-world datasets.
- Performed data preprocessing, exploratory data analysis, feature engineering, and visualization.
- Contributed to developing and evaluating machine learning models.
- Used data-driven insights to support project analysis and decision-making.
Excelerate · Internship
Aug 2025 – Sep 2025
- Performed data cleaning, transformation, and exploratory analysis using SQL, Excel, and PostgreSQL.
- Developed interactive dashboards and reports using Looker Studio.
- Analyzed KPIs and business metrics to identify trends and actionable insights.
- Supported data-driven reporting and visualization for project stakeholders.