Tanuj Saxena

I’m a Software Development Engineer focused on building AI/ML systems that are scalable, efficient, and retrieval-first—especially for NLP, RAG, and distributed architectures. I enjoy turning complex document workflows into dependable automation and measurable performance gains.

In my internship at Cvent, I engineered NLP automation workflows using Python and transformer-based pipelines, improving enterprise content turnaround by 30%. I also developed semantic indexing and retrieval across 10K+ enterprise documents using vector embeddings and optimized query pipelines.

At The AIZoned, I architected scalable retrieval pipelines using Pinecone and transformer embeddings across 50K+ vectors. I built multi-agent asynchronous AI workflows using LangGraph, reducing execution latency by 28%, and fine-tuned LLaMA models using LoRA/QLoRA while streamlining training workflows for efficient GPU utilization.

My projects reflect this same engineering mindset: scalable indexing and structured knowledge mapping for a Legal AI Knowledge Graph, a cloud-native Sanskrit RAG platform using SanskritBERT embeddings and FAISS (with hybrid BM25 + dense search improving retrieval quality by 18%), and a multimodal Virtual Try-On system using DensePose, UNet synthesis, and Stable Diffusion that reduced preprocessing and alignment latency by 22%.