Credit Card Fraud Detection
→
Summary
Developed a robust credit card fraud detection system leveraging advanced machine learning techniques and optimized data preprocessing for high accuracy.
About
Highly analytical and results-driven B.Tech graduate with proven expertise in Data Engineering, Machine Learning, and AI development. Adept at designing and deploying scalable solutions, from RAG-based chatbots to high-accuracy NER systems, leveraging cloud platforms like Azure and advanced ML frameworks. Eager to contribute strong technical skills in NLP, computer vision, and MLOps to innovative data-driven initiatives.
Work
Air India Limited
|Lead Data Engineer
Remote, India
→
Summary
Led the development of a RAG-based AI chatbot for Air India, focusing on optimizing retrieval accuracy and real-time response generation for complex aircraft maintenance queries.
Highlights
Implemented semantic chunking using LangChain's text splitters to preprocess aircraft maintenance manuals into context-aware chunks (512-1024 tokens), significantly optimizing retrieval accuracy for technical queries in Azure AI Search.
Configured Azure AI Search with Azure OpenAI embeddings to enable hybrid search (vector + keyword), improving precision for aircraft system troubleshooting queries by 15% through iterative testing.
Built a serverless backend using Azure Functions and FastAPI, integrating Azure OpenAI (GPT-4) to generate responses grounded in retrieved chunks, reducing latency to <2s per request for real-time user interactions.
Designed robust REST APIs for streaming responses and metadata retrieval, reducing frontend integration time by 20% via Swagger documentation and Postman collections, emphasizing modularity and reusability.
Containerized the backend with Docker and automated CI/CD via Azure DevOps, enabling 5x faster deployments for scalable cloud hosting through optimized image builds and pipeline orchestration.
Classforma
|Machine Learning Intern
Remote, India
→
Summary
Engineered a high-accuracy TensorFlow NER system and built scalable NLP workflows, demonstrating expertise in model development and optimization.
Highlights
Engineered a TensorFlow NER system using BERT/DistilBERT, achieving 98% accuracy in classifying entities (persons, organizations, locations, etc.) via optimized preprocessing workflows.
Researched transformer architectures, enhanced efficiency, and built a scalable Colab pipeline for NLP workflows, streamlining development and experimentation.
Cut inference time by 30% via DistilBERT's lightweight design, while retaining accuracy and thoroughly documenting size-precision trade-offs.
Volunteer
IIT Kanpur
|Organizer, Show Management Udghosh'22
Kanpur, Uttar Pradesh, India
→
Summary
Led show management for Udghosh'22, a major college fest at IIT Kanpur, overseeing logistics, design, and administrative coordination.
Highlights
Guided a 2-tier team of approximately 30 senior executives, effectively managing the show management wing of Udghosh.
Designed prototypes for artworks that were subsequently mass-produced by a 2-tier team, ensuring consistent visual branding.
Successfully managed and secured necessary permissions from college authorities, including the estate office and computer center, to ensure smooth execution of the fest.
Education
Indian Institute of Technology Kanpur
Kanpur, Uttar Pradesh, India
→
Bachelor of Technology
Aerospace Engineering
Grade: 6.8/10
Courses
Data Structures and Algorithms
Fundamentals of Computing
Partial Differential Equations
Operating Systems
Linear Algebra and Ordinary Differential Equation
Software Development and Operations
Rani Laxmi Bai School, Lucknow
Lucknow, Uttar Pradesh, India
→
Senior Secondary (CBSE - XII)
Science
Grade: 82.8%
Rani Laxmi Bai School, Lucknow
Lucknow, Uttar Pradesh, India
→
Secondary (CBSE - X)
All Subjects
Grade: 10.0/10.0
Certificates
Node.js, Express, MongoDB & More: The Complete Bootcamp
Issued By
Udemy
Amazon ML Summer School 2022
Issued By
Amazon India
Skills
Cloud Platforms & Tools
Azure, CI/CD, REST APIs, Postman, Swagger, Colab.
Programming Languages
C, C++, HTML, LaTeX, Python, JavaScript, PostgreSQL.
Software & Libraries
Express.js, MongoDB, Node.js, LangChain, NumPy, OpenCV, TensorFlow, Keras, BERT, DistilBERT, Scikit-learn, Docker, Azure DevOps, FastAPI, Azure Functions, Azure OpenAI, Azure AI Search.
Machine Learning & AI
Natural Language Processing (NLP), Computer Vision, Deep Learning (DNN), Supervised Learning, Unsupervised Learning, Model Evaluation, Fraud Detection, Object Detection, Named Entity Recognition (NER), PCA, SVM, Random Forest, MLP, Canny Edge Detector, Hough Transformation, Histogram of Oriented Gradients (HOG).
Projects
Computer Vision: A Classic and Modern Approach
→
Summary
Explored and implemented both classic and modern computer vision algorithms for tasks such as edge detection, lane detection, object detection, and digit classification.