Rishabh Kumar

Dedicated software engineer with a strong passion for AI, deep learning, and full-stack development. Experienced in solving real-world problems using cutting-edge technology.

Chennai, Tamil Nadu, India

LinkedInX
RK

About

Experienced in building AI-driven solutions and modern web applications. My expertise lies in developing efficient, scalable platforms using TypeScript, Node.js, React, and more. I focus on delivering impactful solutions that streamline processes and improve user experiences.

Work Experience

Hotelzify
On-site

2024 Sept - Present

AI Intern

  • Engineered a comprehensive management dashboard for the Qdrant vector database, enabling real-time monitoring, querying, and in-place updating of knowledge base chunks. This system featured automated duplicate detection and removal, streamlining KB maintenance and eliminating the need for direct API interaction.
  • Architected and implemented an ultra-low latency Retrieval-Augmented Generation (RAG) pipeline, achieving sub-millisecond vector retrieval speeds to deliver highly accurate, real-time semantic search results.
  • Developed a sophisticated AI voice interface built upon an agentic framework with multi-tool integration, empowering the system to autonomously perform complex actions such as processing bookings, verifying availability, and distributing media.
  • Designed a highly optimized, node-based query system for internal hotelier operations, structuring complex tasks into an agentic workflow to ensure minimal latency and high throughput.
  • Deployed and managed high-performance, containerized APIs on AWS, engineered to handle high-concurrency requests with robust, persistent session state management for seamless conversational continuity.

Marki
Remote
USA, LA

2024 Jan - 2024 Aug

Full Stack Developer

  • Architected and led the end-to-end development of a bespoke meal ordering platform, utilizing Next.js, TypeScript, and Node.js to create a scalable, high-performance solution that streamlined order processing and meal prep customization.
  • Engineered a low-latency, real-time data integration with the Uber Eats API, fetching comprehensive order details to power a sophisticated recommendation engine that provided personalized meal suggestions based on users' blood work analysis and fitness goals.
  • Implemented a high-throughput data processing pipeline that integrated the USDA food database, enabling real-time nutritional analysis and validation for all meal components to ensure dietary accuracy.
  • Deployed and managed the full-stack application on AWS, utilizing Docker for containerization and establishing a CI/CD pipeline to ensure automated, zero-downtime deployments and high availability.

MadgenAI
Remote
USA, SF

2023 Nov - 2024 Jan

AI Engineer

  • Architected and developed a high-performance media generation platform by integrating state-of-the-art diffusion and transformer models. I engineered complex, node-based generative workflows using ComfyUI, enabling precise control over the content creation pipeline from model loading to final output.
  • Deployed the generative workflows as a scalable, production-ready API service, optimizing for high-concurrency and efficient resource management. This involved containerizing the ComfyUI environment and leveraging cloud infrastructure to handle asynchronous job queuing and deliver a robust, enterprise-grade generative AI service.

Education

Vellore Institute of Technology

2022 - 2026
Bachelor's Degree in Electronics and Computer Engineering
Relevant Courses: DSA, AI and Machine Learning, Deep learning, Discrete Mathematics, Complex Numbers, Computer networks, Probability and Statistics

Skills

Langchain
LLAMA Index
Livekit
Speech-to-Text
Text-to-Speech
Voice AI
Real-time Processing
WebRTC
Audio Processing
TypeScript
Next.js
Node.js
Python
AI/ML
React
JavaScript

Projects

CogniDB: Natural Language Database Querying

  • Authored and published CogniDB, a Python library downloaded over 5,000 times, by designing a modular driver system that abstracts the complexities of MySQL, PostgreSQL, and MongoDB into a single, unified interface.
  • Implemented a zero-trust security layer to mitigate LLM-generated threats, creating a custom SQL sanitizer that strictly enforces read-only operations to make the library safe for production use.

Python
MySQL
PostgreSQL
MongoDB
Natural Language Processing

FreshOut: Conversational E-Commerce Platform

  • Developed a sophisticated context-aware system that injects the application's real-time state—including UI interactions and cart contents—into the AI's prompt to enable precise, human-like command interpretation.
  • Solved the complex challenge of cross-tab state synchronization by implementing a dual-layer persistence architecture using IndexedDB for session recovery and the localStorage API for real-time updates.

Next.js
TypeScript
AI Integration
IndexedDB

Gmail & Calendar AI Assistant

  • Architected a multi-agent system using the Mastra framework to solve the problem of single-threaded conversational context, allowing the bot to handle complex, chained commands like 'find a meeting time in this email and schedule it.'
  • Integrated the assistant directly with Google Workspace APIs, managing the OAuth authentication flow and token lifecycle to create a unified, secure, and powerful conversational interface.

Mastra Framework
Google Workspace APIs
OAuth
Multi-Agent Systems

AI-Powered Product Recommendation System

  • Developed a recommendation engine that leverages NLP to interpret the semantic intent behind user queries, overcoming the limitations of traditional keyword-based matching to deliver more accurate product suggestions.
  • Deployed the NLP engine within a conversational chat interface, creating a seamless and natural shopping experience that outperformed standard industry recommendation tools.

Natural Language Processing
Artificial Intelligence
Database Management System

ChatCSV: AI-powered Data Analysis Platform (SaaS)

  • Architected a hybrid AI system that delegates complex data analysis from the LLM to a secure, sandboxed Python runtime, overcoming inherent LLM limitations in mathematical accuracy to ensure factual, reproducible insights.
  • Built and deployed the full-stack application on AWS, creating a user-friendly interface for file uploads and interactive visualizations while ensuring the backend could scale to handle high-concurrency analysis jobs.

Full Stack Development
Artificial Intelligence
Data Visualization

LawIQ - Legal Chatbot with Real-time Voice Interaction

  • Solved the challenge of domain-specific language by fine-tuning NLP models on legal terminology, significantly improving the chatbot's ability to accurately comprehend and respond to specialized legal queries.
  • Architected the system to support real-time voice interaction, creating an accessible and intuitive interface that automates routine legal inquiries and improves access to foundational legal knowledge.

NextJs
Natural Language Processing
Artificial Intelligence

VocalGPT: Terminal-Based AI Voice Chat

  • Built a lightweight, high-performance voice chat simulator in Python, specifically optimized for low-latency audio stream processing directly within a terminal environment.
  • Designed the tool for practical applications like phone call simulations and interactive language practice, providing developers with a simple yet powerful tool for building real-time conversational AI.

Python
Large Language Models
Artificial Intelligence

Solana Multi-Wallet Manager

  • Constructed a high-throughput transaction processor in Rust to leverage its memory safety guarantees and fearless concurrency, building a stable and performant system for the Solana ecosystem.
  • Optimized system reliability by implementing parallel transaction signing and broadcasting, increasing efficiency by 3x and achieving a 99.9% success rate through robust error handling and automated retry logic.

Rust
Solana SDK

OneTicket - AI-Powered Ticket Booking Chatbot

  • Designed a fully conversational NLP flow that completely abstracted the booking process, allowing users to secure tickets through natural language without interacting with a single traditional UI element.
  • Implemented an automated QR code generation and delivery system to streamline event entry, contributing to the project's success in ranking 8th out of 600 teams at the Smart India Hackathon.

Natural Language Processing
AI/ML
Chatbot Development