Engineering intelligence through scalable MLOps and production-grade deep learning systems.
“The important thing is not to stop questioning. Curiosity has its own reason for existing.”
— Albert EinsteinSoftware • ML • AI
Software engineer transitioning into ML/AI engineering, combining production full-stack experience with applied work in RAG, computer vision, and reproducible ML pipelines.
I build end-to-end software and ML applications, from data preparation and model evaluation to APIs, testing, and deployment. My work combines practical machine learning with production software engineering to turn working prototypes into reliable applications.
Building RAG, NLP, and computer-vision projects with measurable evaluation and practical user workflows.
Creating tested ML pipelines with DVC, MLflow, GitHub Actions, FastAPI, and AWS-backed experiment artifacts.
Art Evo
Designed and delivered a production e-commerce platform as the sole developer, covering customer workflows, seller operations, payments, and deployment.
Image Express, Ireland
Developed and maintained an e-commerce platform for digital photo printing and custom gifts as part of a four-person engineering team.
Lince Soft Pvt. Ltd.
Contributed to frontend and backend development while building foundational software engineering experience.
Northumbria University, London
Advanced postgraduate study in computing and information science, strengthening software development, system design, and applied technology skills.
CVR College of Engineering, India
Built a strong engineering foundation in analytical reasoning, quantitative problem-solving, and structured system design.
Applied ML, AI, and software-engineering projects covering RAG, NLP, computer vision, evaluation, reproducible pipelines, and production full-stack development.
Document Q&A application supporting PDF, Markdown, and TXT uploads, with sentence-aware chunking, vector retrieval, source traceability, and evaluated retrieval performance.
Seven-stage reproducible ML pipeline using DVC, TF-IDF, and model comparison. Achieved 90.36% accuracy with Linear SVM across 39,665 reviews.
ViT-based face-matching system that generates image embeddings and compares faces using cosine similarity, mean-pooled reference embeddings, and configurable thresholds.
CNN image classifier built from scratch with TensorFlow/Keras, data augmentation, and separate training, validation, and test sets, achieving 80.24% test accuracy.
Production e-commerce platform for 3D-printed art and decor, featuring authentication, cart, wishlist, checkout, seller operations, Razorpay payments, and live order tracking.
Initiate a connection to discuss Neural Architectures, MLOps, or deployment strategies. Or just want to say Hi :) .