Shreyas Varadkar
AI Engineer
About Me
Hi, I'm Shreyas Varadkar, an AI Engineer focused on building RAG systems, LLM automation workflows, MCP-based integrations, and voice-enabled conversational AI.
I work on creating intelligent agents that combine language models with structured knowledge retrieval, vector databases, and real-time speech capabilities. My experience includes open-source LLM exploration, backend AI service development, and building scalable, production-ready AI pipelines.
Selected among the top 3000 students nationwide for the prestigious Amazon ML Summer School 2023.
I enjoy solving real-world problems with practical AI, experimenting with new models and techniques, and continuously improving as an engineer.
Work Experience
Jr. SDE(AI)
Amplispot, Mumbai
Led AI initiatives at Amplispot, building automation workflows, RAG-based systems, and voice-enabled AI products to optimize internal and client-facing workflows.
- Designed and deployed an AI Service Framework in NestJS with Keycloak-based authentication and Kubernetes deployment, improving scalability and security across internal AI tools.
- Developed RAG chatbots and conversational AI systems, reducing hallucinations by ~40% and improving contextual accuracy.
- Built a bilingual (English & Hindi) voice-based conversational AI bot for the insurance domain using ElevenLabs, Express, and HTML, enabling interactive voice journeys and personalized plan suggestions.
- Built AI-assisted template automation using OpenCV and LLMs to convert design images into Conversify-compatible JSON templates, reducing a manual 1-2 hour workflow to seconds.
Junior Data Scientist
Zummit Infolabs, Bangalore
- Worked on migrating ML workflows of an "HR interview bot" project enhancing automated detection of candidate's nervousness and confidence based on displayed emotions.
- Led the Third Eye project, which aims to build AI-powered spectacles to assist visually impaired people where I optimized object detection using YoloV8, achieving a 20% accuracy boost.
ML Trainee - Amazon ML Summer School
Amazon
- Secured a position among the top 3000 nationwide, gaining invaluable exposure to state-of-the-art Machine Learning applications.
- Engaged in intensive workshops and training sessions conducted by accomplished data scientists currently working at Amazon.
Data Science Intern
Codeclause, Pune
- Developed a Machine Learning powered solution for predicting Churn in Telecom Industry to identify patterns and various trends that influence customer satisfaction and helps to reduce churn in telecom.
Internship Trainee
Marine Industrial Control & Automation Labs
- Acquired comprehensive training in Python Programming and MySQL Database, skillfully applying them in a collaborative group project titled 'Database Queries'.
Skills & Technologies
Programming Languages
Python
Java
JavaScript
Backend & Frameworks
NodeJS
Express
NestJS
AI & Machine Learning
AI/ML
Ollama
MCP
Databases
PostgreSQL
MySQL
MongoDB
DevOps & Cloud
Docker
AWS
Git
Tools & Others
n8n
Postman
Power BI
Featured Projects
MyAptitude - Online Aptitude Exam System
A web application designed to conduct online aptitude exams for companies aiming to eradicate paper leaks through real-time question scraping and customization.
LangGraph based Mutual Fund Q&A Chatbot
A multi-document RAG chatbot for answering HDFC Mutual Fund queries across monthly factsheets, with support for follow-up questions and general conversation.
VoxRefine - Local Voice Transcription App
A fully local transcription app that turns voice input into cleaned, searchable transcripts using faster-whisper and a local LLM, with zero cloud dependencies.
Smart ATS Resume Analyzer using AI
An AI-powered resume analyzer that compares resumes against job descriptions, highlights missing keywords, and helps improve ATS compatibility.
Get In Touch
I'm currently looking for new opportunities. Whether you have a question or just want to say hi, feel free to reach out!
Note: You can type in the details here, but unfortunately this won't work. But don't worry if I build your form that will work. For actual contact, please use the links below.
Let's Connect
