Prabhat Kumar
I train models, build products, and ship them to production.
AI engineer at Rakuten India, building LLM-powered automation, RAG pipelines and full-stack tools in production. Research published at an international Springer conference in Singapore.
Prabhat Kumar
AI/ML · Full Stack · DevOps
Technologies I Work With
About Me
Passionate about building intelligent, scalable systems
const prabhat = { name: "Prabhat Kumar", role: "AI/ML + Full Stack", focus: ["LLMs", "RAG", "FastAPI"], cgpa: 9.4, publication: "Springer, Singapore", openTo: "Full-time roles", }
Currently at Rakuten India
I work on production AI systems at Rakuten India — LLM automation, RAG pipelines, vector search and full-stack engineering. I graduated in Computer Science (Data Science) from RN Shetty Institute of Technology under VTU with a 9.4 CGPA. My work spans the full stack: from training and fine-tuning language models to shipping them behind FastAPI services with Docker and CI/CD. I also build React and Next.js frontends and automate infrastructure with shell scripting and Kubernetes.
Currently working at Rakuten India, building LLM-powered automation, RAG pipelines and full-stack internal tools. Published research at an international Springer conference in Singapore.
GitHub Repos
LeetCode Solved
GitHub Followers
CGPA
Experience
My professional journey so far
AI/ML Engineering Intern
Rakuten India
Jan 2026 – Present · Bengaluru, India
- Engineered production-grade RAG pipelines using embeddings, vector search and reranking, improving retrieval precision for Rakuten's enterprise knowledge bases
- Developed LLM-powered automation workflows that process large-scale structured and unstructured data for Rakuten's internal tooling
- Built full-stack internal tools with React, TypeScript and FastAPI as part of the CIS Autoconfig infrastructure configuration project
- Applied prompt engineering and context-window optimisation to reduce LLM response latency and token costs
Data Science Intern
Cognify Technology
Jun 2025 – Aug 2025 · Bengaluru, India
- Designed interactive analytics dashboards to surface key business metrics for cross-functional stakeholders, improving data visibility across teams
- Trained and evaluated ML classification models for user behaviour prediction using scikit-learn, achieving strong accuracy on held-out test sets
- Automated ETL data extraction pipelines using Python, reducing manual processing time by 40 percent and eliminating recurring data-quality errors
Student Placement Coordinator
RN Shetty Institute of Technology
Jan 2024 – Mar 2025 · Bengaluru, India
- Led campus recruitment drives in collaboration with 10+ technology companies, facilitating 200+ student placements
- Mentored peers through resume reviews, mock interviews, and placement-readiness workshops
- Coordinated pre-placement talks, aptitude tests, and GD/PI rounds, ensuring smooth end-to-end candidate experience
- Managed placement records, student–company communications, and onboarding documentation throughout the hiring cycle
Event Organiser
IEEE Student Branch
Jan 2024 – Dec 2024 · Bengaluru, India
- Planned and executed IEEE community events focused on emerging technologies and professional development
- Coordinated logistics, speakers, and attendee engagement, growing branch participation over the year
Skills
Tools and technologies I use to build, train and ship AI-powered products
AI/ML
Full Stack
DevOps & Cloud
Languages
Data & Viz
Projects
Case studies and open-source work
Featured Projects
Medical RAG Chatbot
Making medical knowledge searchable in plain language
A retrieval-augmented chatbot that answers medical questions by searching a curated knowledge base, built to cut the time it takes to look up clinical information.
LLM Fine-Tuning Pipeline
Training domain-specific language models end-to-end
A complete pipeline for fine-tuning open-source LLMs with QLoRA, covering dataset preparation, training runs, evaluation metrics and model export.
Grocery Full-Stack App
A complete e-commerce platform from UI to deployment
A full-stack grocery ordering app with product listings, cart management, user auth and Docker-based deployment, covering both frontend and backend from scratch.
All Repositories
LLM Fine Tuning
Fine tuned a qwen llm model on my own data on google colab
BASH-Scripting-linux-zsh-
creating a game in using scripting in .sh file
Grocery Full stack web app
using python flask and sqlite,html,css,javascript(full stack web application)
Gen-AI-Complete
Complete Generative AI including ml,nlp,dl,langchain and huggingface,etc.
ANN Classification Churn-bank-
No description available
Medical Chatbot
Medical assistant with locally using llm to give response and using the database mongodb with new technology next js
VR-System-to-Simulate-Soft-Hand-off-among-Towers-in-Mobile-Network
Our project, VR System to Simulate Soft Hand-off among Towers in Mobile Networks, aimed to create an immersive simulation of mobile network handoff between towers. A team of five members worked collaboratively to model a 3D representation of a city, complete with mobile towers (base stations) and cars driving at high speeds.
SQL FULL Prep
This repo contains leetcode,hackerrank,stratascratch sqls question answer .
blog-platform-springboot-react
Engineered a secure full-stack blog platform with Spring Boot and React, featuring JWT authentication, REST APIs, and role-based access control.
Spring Spring boot
complete spring framework and springboot with all JWT and Oauth
Gesture Based Calculation Comuter vision
Gesture-Based Numeric Drawing and Real-Time Calculation System This project leverages AI and computer vision to create an intuitive gesture-based interface for numeric input and real-time calculations
Ads CTR Optimizer-Reinforcement-Learning-
Ads_CTR_Optimizer-Reinforcement-Learning ,to determine which ad is optimial by recieving the feedback.
SmartTrend-ARL
The main objective is to identify item relationships in transactional data. For example: - If a customer buys **bread**, they are likely to buy **butter**.
Unity Game-Asset-Management-System
I developed a Unity Asset Management System that combines Unity for game development, PHP for server-side operations, and MySQL for database management. The system allows users to log in, and their details, such as username and score, are securely updated in the MySQL database. C# is used within Unity to handle in-game operations.
Insta BOT
"I have developed an Instagram bot using Python and the Instagram API, designed to perform a variety of tasks such as automating posts, liking, commenting, following, and unfollowing. This bot enhances user engagement and streamlines social media management by automating routine interactions, ultimately saving time and improving efficiency."
OCR Documet Reader
This OCR project leverages Python to extract text from images, enabling the conversion of printed or handwritten documents into machine-readable text. By combining advanced libraries and tools, the system processes image data efficiently, recognizing and translating characters with accuracy.
Vox-Pilot
This virtual assistant is built using Python and utilizes speech recognition to understand voice commands. It can perform various operations, such as opening YouTube or other applications, based on user input. By leveraging Python's powerful libraries, the assistant efficiently translates spoken words into actions, providing a seamless experience.
Linux-System-Monitor-Script
A python script to monitor the system for linux os
Blog LLM AI DATA
This is the full stack web app, personal blog website created using python Flask
Salary Prediction
Predicting the salary of employee based on the years of Experience using simple linear regression
Sentiment-Analysis Restaurant-NLP-
Review the restaurant using Natural Language Processing
Churn Detection Bank Customer
This project focuses on predicting **customer churn** in the banking sector using an **Artificial Neural Network (ANN)** built from scratch. By leveraging historical customer data, the model identifies patterns that help predict whether a customer is likely to leave the bank or remain loyal.
CNN classification Cat or dog
This project focuses on classifying images of cats and dogs using a **Convolutional Neural Network (CNN)** built from scratch
ANN-model-to-predict-the-electrical-energy-
Trained ANN to predict the Electrical energy
ChatBot-Frontend
frontend for the chatbot project so that we can host it using the kubernetes
Competitive Programming
LeetCode & HackerRank achievements
70.37%
Acceptance
#311,470
Ranking
Difficulty Breakdown
Easy
Medium
Hard
Achievements
Publications, certifications, and milestones
Publications
Real-Time Gesture Recognition for Numeric Drawing and Calculation
5th International Conference on 6G Communications Networking and Signal Processing (SGCNSP)
Springer · Nanyang Technological University, Singapore · Dec 2025
Certifications
Microsoft Certified: Azure Data Fundamentals
Microsoft
Deep Learning (Elite)
NPTEL - IIT Ropar
Data Science For Engineers
NPTEL
AWS AI and ML Certifications
Amazon Web Services
DevOps Foundation
Great Learning
Docker & Kubernetes
Udemy
AI Agent Fundamentals
Databricks Academy
Education
Graduated 2026 with a 9.4 CGPA
B.E., Computer Science (Data Science)
RN Shetty Institute of Technology
VTU (Visvesvaraya Technological University) · Bengaluru, India
Get In Touch
Have a project in mind or want to collaborate? I'd love to hear from you.