Devyanshi Bansal

I am an undergraduate at Delhi Technological University (DTU) with a passion for AI and Machine Learning.

What I Have Built
Here are some of the key projects I have worked and contributed to:

1. INDRA: AI for Water Sustainability
   A GIS-driven system for assessing rainwater harvesting feasibility.

Tech Stack: Neo4j (Graph DB), Qdrant (Vector DB), RAG, Computer Vision.

Key Feature: Uses Graph RAG to provide water-based crop recommendations and automates tank sizing with 14.8 cm³ precision.

Impact: Reduces manual assessment effort by 70% for rural users.

2. IntervueAI: Automated Interviewer
   An AI-powered technical interviewing and proctoring platform.

Tech Stack: CodeBERT, Knowledge Graphs, Computer Vision, Speech-to-Text.

Key Feature: Real-time difficulty adaptation and semantic grading consistency (improved by 35%).

Capabilities: Anomaly detection within 500ms and 92% transcription accuracy.

3. Qloo Fashion AI (Hackathon Project)
   A cultural fashion recommendation system built for the Qloo Hackathon.

Tech Stack: FastAPI, Qloo API, Gemini AI, Qdrant.

Features: Natural language search, cultural style fusion, and "anti-recommendations" for unique fashion discovery.

4. AI Finance Chatbot
   An intelligent conversational assistant for financial insights and queries.

Tech Stack: NLP, Python, Generative AI.

Features: Provides real-time financial assistance and interactive query resolution ( You can add specific features here like "stock tracking" or "budget advice").

5. ML Loan Predictor
   An end-to-end machine learning pipeline for financial risk assessment.

Tech Stack: Python, XGBoost, Ensemble Models, Pandas.

Performance: Achieved 94% accuracy and a Mean Absolute Error (MAE) of 0.35 on the Kaggle dataset.

Skills I Possess
Languages: Python, C++, Java, SQL, HTML/CSS.

AI & ML: TensorFlow, PyTorch, Scikit-Learn, LangChain, HuggingFace.

Databases & Retrieval: Qdrant (Vector DB), Neo4j (Graph DB), Pandas.

Web & Tools: FastAPI, Docker, GitHub, Linux, Figma.

Core Competencies: RAG (Retrieval-Augmented Generation), Knowledge Graphs, Computer Vision, NLP.

What I Aim to Learn:
I am currently pushing the boundaries of Retrieval-Augmented Generation (RAG) and moving toward Agentic AI. My next learning milestones include:

Agentic Workflows: Mastering frameworks like LangGraph to build autonomous agents that can plan and execute multi-step tasks.

Advanced MLOps: Learning to deploy and scale Large Language Models (LLMs) in production environments efficiently.

Graph Neural Networks (GNNs): Deepening my understanding of Graph AI to build even more robust recommendation and analysis engines.