Sujal Maheshwari

AI Architect at STARTHACK.IO (Freelance)

Built low-latency AI agents for automated code evaluation using parallel execution, reducing latency from 7 minutes to 30 seconds. Architected a real-time voice-to-voice technical interview system using GPT Real-Time APIs with dynamic persona switching. Designed scalable code chunking and repository indexing strategies, improving RAG retrieval accuracy from ˜ 62% to ˜ 86%.

AI Data Scientist Intern at Basal AI

Implemented parallel, location-wise Shopify data extraction via web scraping, reducing product and variant compilation time from 5+ hours to under 45 minutes for 130K+ Products and variants. Built a fine-tuned GPT-4.1 content generation pipeline, reducing content turnaround from 3 days to 15 minutes with 96% baseline accuracy (Azure AI Foundry). Implemented CI/CD workflows with GitHub Actions for automated testing and deployment of AI pipelines.

LLM Engineer at Namaste Enabled Service Robot (Patent Work)

Developed an NLP-based keyword extraction chatbot using rule-based and statistical filtering techniques. Fine-tuned GPT-2 on proprietary hospitality-industry data to improve conversational relevance. Implemented a Retrieval-Augmented Generation pipeline for document-grounded responses.

AI Architect at STARTHACK.IO

Led the development of low-latency AI agents for automated code evaluation using parallel execution, reducing system latency from 7 minutes to 30 seconds. Architected a real-time voice-to-voice technical interview system leveraging GPT Real-Time APIs with dynamic persona switching. Designed scalable code chunking and repository indexing strategies, boosting RAG retrieval accuracy from approximately 62% to 86%.