Full Stack Developer
ASSANJ
May 2026 - Present
- Developing and optimizing full-stack web applications using React, Node.js, and REST APIs; collaborating in an agile, industry-oriented environment to ship features across the complete development lifecycle.
- Implementing scalable backend logic, responsive UI components, and streamlined CI/CD workflows for real-world client-facing products; contributing to code reviews and architectural decisions.
Backend Developer
Cestrum
Jan 2026 - March 2026
- Building UniTalks — an anonymous real-time chat platform with voice/video via WebRTC, designed to scale to 1,000+ concurrent connections.
- Owned backend architecture around Node.js, Socket.IO, and Redis Pub/Sub for distributed session state; containerized services with Docker and deployed behind Nginx on AWS EC2.
- Set up GitHub Actions CI/CD to cut deployment time by ~75%; load-tested the stack at 150+ concurrent users to validate performance under real traffic.
Generative AI Intern
Analytx4t
Dec 2025 - Feb 2026
- Built production GenAI applications using LLMs, LangChain, and Python; delivered 8 RAG pipelines for document intelligence with vector search and tool-augmented agents.
- Reduced inference latency by 40% via chunking and embedding tuning; evaluated 5+ prompt strategies, boosting RAG faithfulness scores by 22% on internal benchmarks.
Published Patent
System and Method for Stress and Pain Detection Using Multi-Scale Transformer-Based Neural Networks — Application No. 202641065281
RAG Chatbot
Document-grounded Retrieval-Augmented Generation pipelines with LangChain, vector search, and LLMs — including production RAG work that cut inference latency by ~40%.
Ongoing Project — ResearchPilot
AI-powered research intelligence platform combining fine-tuned LLMs, Retrieval-Augmented Generation (RAG), semantic search, and machine learning for paper analysis, abstract enhancement, statistical recommendations, and research gap discovery.
Contact Form
Please contact me directly at dwivedivaibhav3110(at)gmail.com or drop your info here.