Aman Kumar

Full Stack AI Engineer with expertise in building AI-native applications, LLM pipelines, and scalable web solutions.

Chennai, Tamil Nadu

Email
Phone
GitHub
LinkedInLinkedIn
Portfolio
Codolio
AK

About

I am an aspiring Full Stack AI Engineer with a strong focus on building AI-native applications using LLMs, RAG pipelines, and agentic workflows. I have hands-on experience in backend development, full-stack projects, and building production-ready systems that solve real-world problems. My projects span across real-time web applications, AI-driven systems, and conversational AI, showcasing my ability to deliver end-to-end solutions. Alongside development, I have demonstrated excellence in competitive exams and problem-solving: Secured position in the Top 1% of candidates in the National Defence Academy Examination 2021 (UPSC) Ranked among the Top 10% qualifiers of the JEE Advanced 2022 Qualified Round 1 & Round 2 of the Dark Pattern Buster Hackathon 2024, conducted by the Ministry of Consumer Affairs Solved 450+ medium-to-hard coding problems focused on DSA, algorithms, and system design across platforms I enjoy solving problems, building scalable solutions, and constantly exploring new technologies to create meaningful impact.

Work Experience

Epic Stone Media
Remote
Freelance

Nov 2024 - Nov 2025

Backend & Cloud Engineering Intern

Built and deployed cloud-native backend services using Node.js, Express.js, and AWS EC2. Worked on RESTful APIs and microservices, improving system performance and reliability by 30%. Implemented CI/CD pipelines and automated workflows. Designed scalable data layers using PostgreSQL and Redis.

ACM VIT Chennai
Leadership

Apr 2024 - Apr 2025

Treasurer

Managed chapter finances, including budgeting, expense tracking, and allocation of funds. Coordinated with sponsors and stakeholders to secure funding. Ensured transparent financial reporting and compliance with institutional guidelines.

Education

Vellore Institute of Technology

2022 - Present
Bachelor of Technology - 8.68/10

Central Board of Secondary Education

2019 - 2021
Senior Secondary - 75.4%

Central Board of Secondary Education

2019 - 2019
Secondary - 89.8%

Skills

Python
JavaScript
TypeScript
Node.js
Next.js
React
FastAPI
Express.js
PostgreSQL
MongoDB
Redis
AWS
Docker
Git
Prompt Engineering
RAG Pipelines
LLM Integration
Conversational AI

Projects

Privacy-Preserving Edge-Deployed RAG for Clinical Diagnosis

  • Proposed a local-first AI architecture for real-time differential diagnosis using edge-deployed RAG.
  • Designed a multi-stage AI system integrating ASR, NER, and FAISS-backed retrieval.
  • Achieved 94.1% Top-3 diagnostic accuracy and 1.2s latency on 500 clinical cases.
RAG
ASR (Whisper)
FAISS
LLaMA
Python

AttendEase – Smart Library Management System

  • Developed a full-stack library seat booking and attendance management system with real-time analytics and QR code check-in/out
  • Implemented Node.js, MongoDB, and JWT for backend logic, authentication, and secure data flow
  • Features include dynamic seat reservation, time-slot based booking with conflict detection, QR-based attendance, and admin dashboard
  • Added intelligent analytics: peak hour usage, occupancy tracking, and weekly/monthly utilization reports
Node.js
MongoDB
JWT
Full Stack

MockMate AI – Voice-Based Interview Preparation

  • Designed an AI-native interview platform using voice agents for real-time mock interviews.
  • Engineered an LLM-integrated workflow optimizing prompt design.
  • Built a scalable full-stack system with authentication and session tracking.
Vapi AI
Gemini
LLM
Full Stack

Text2Query – Natural Language to SQL

  • Developed a NL2SQL system supporting MySQL, PostgreSQL, and MongoDB with LLM-based query generation.
  • Designed semantic caching mechanisms reducing LLM inference cost by 40%.
  • Built a prompt optimization and validation pipeline using constitutional AI to prevent SQL injection.
LLM
Prompt Engineering
MySQL
PostgreSQL
MongoDB

AI Chatbot-Based Ticketing System

  • Built a multilingual AI chatbot system using agent-based workflows to automate booking.
  • Designed end-to-end conversational pipelines with intent detection and tool calling.
  • Improved system reliability through structured workflows and validation logic.
Python
Conversational AI
Agents
Full Stack

FreshOut – AI Shopping Assistant

  • Built an interactive e-commerce platform enabling natural language commands for browsing, cart management, and navigation
  • Implemented a context-aware shopping assistant that understands current page, cart contents, and hover interactions
  • Engineered cross-tab synchronization and persistent sessions using IndexedDB and localStorage events
  • Designed point-and-talk interaction where the assistant recognizes which product the user is viewing
Next.js
React
IndexedDB
LLM Integration

CollabEditor — With Realtime Notifications

  • Built a modern collaborative document editor supporting real-time editing, live notifications, comments, and user presence
  • Integrated authentication (Clerk), error monitoring (Sentry), and rich text editing (Lexical, JSM Editor)
  • Utilized Technologies: Next.js, React, Tailwind CSS, Liveblocks, Clerk, Sentry, TypeScript
Next.js
React
Tailwind CSS
TypeScript
Clerk
Sentry

OneTicket

  • Developed a multi-agent system to simplify museum ticket booking by enabling natural language input
  • Ranked top 10 out of 600 participating teams in the Internal Smart India Hackathon
  • Implemented with LangChain, Next.js, and the VercelAI SDK
LangChain
Next.js
VercelAI SDK

Gmail-n-Calendar Bot

  • Built a Telegram bot to manage Gmail and Google Calendar through chat interface
  • Implemented email management (read, send, search) and integrated calendar control (list, create, update events)
  • Designed an intelligent Email coordination agent to automatically schedule meetings from requests found in emails
  • Developed using the Mastra framework, Google API libraries, Node.js, and TypeScript in a multi-agent architecture
Node.js
TypeScript
Google APIs
Mastra Framework
Telegram Bot API

Shopify Scrapper – Mobile App

  • Built a production-ready monorepo with a Fastify/TypeScript backend and Expo React Native frontend for Shopify store data extraction
  • Implemented user authentication, credit-based purchasing, and structured CSV result downloads
  • Integrated Supabase for auth and data management with environment-based configuration for local and production
  • Added CORS, rate limiting, and secure API routes for authentication, scraping, and payment flows
Expo
React Native
Fastify
TypeScript
Supabase

File Sharing API – Secure Backend Service

  • Developed a scalable backend service for secure file management with Express, PostgreSQL, Amazon S3, and Redis
  • Implemented JWT-based authentication with HTTP-only cookies and secure file upload/retrieval via presigned S3 URLs
  • Added Redis caching for file listings and search results to reduce database load and improve performance
  • Containerized all services with Docker for independent deployment and scalability
Express
PostgreSQL
Amazon S3
Redis
JWT
Docker

Resource Allocation in Cloud Computing

  • Developed an innovative resource allocation system for cloud computing using game theory principles
  • Implemented a Nash equilibrium-based algorithm to optimize resource distribution among multiple cloud users
  • Achieved improved resource utilization and fair allocation through mathematical modeling and simulation
Game Theory
Python
Cloud Computing
Optimization

E-Commerce Profitability Analysis

  • Analyzed product data (pricing, revenue, ratings, reviews) to identify key profitability drivers for an online sports retailer
  • Used SQL and Python (Pandas, NumPy) for data cleaning, pricing analysis, and performance evaluation
  • Provided insights on price labeling, discount optimization, and sales strategy
SQL
Python
Pandas
NumPy