AI Intern
Summary
Developed a complex Graph RAG architecture for social media comment moderation, resulting in significant improvements in accuracy and reducing manual review time by 40%, enhancing overall operational efficiency.
Highlights
Tech-Stack: Langgraph,, Python, FAST-API, Docker, Prometheus, MySQL, Kubernetes
Skills: GenAI, RAG, NLP, Backend Development, GCP, BigQuery
DataScience -SDE Intern
Summary
Engineered a Retrieval-Augmented Generation (RAG) system for cybersecurity compliance reviews, achieving 95% accuracy after fine-tuning the LlamaIndex-7b LLM model, thus improving review efficiency by 60%.
Highlights
Tech-Stack: Langchain, LlamIndex-7b, Python, Transformers, Ragas
Skills: Machine Learning, NLP, Analytics, Google Cloud
AI Engineering Intern
Summary
Contributed to the content generation pipeline by integrating various diffusion models and engineered an SDK for AI services which served as the backbone of AI services. Set up the backend pipeline for image storage and retrieval from cloud storage, reducing time by 40%.
Highlights
Tech-Stack: Tensorflow, Pytorch, Numpy, SQL, VertexAI
Skills: Machine Learning, Programming, Data Analysis, GenerativeAI
Research Intern
Summary
Machine Learning and NLP: Developed a NER-based dataset extractor from research papers. Integrated the GROBID server services in the code for file conversion.
Highlights
Tech-Stack:Python, GROBID, NLP
Codechef Profile☑ Leetcode Profile
Qualified stage I, was among top 1 percentile candidates at state level.
Qualified stage I, was among the top 1 percentile candidates at the national level.
C, C++, Python.
Pytorch, Docker, BigQuery, VertexAI, Langgraph, Langchain, FAST API, Google Cloud.
Summary
A LLM chatbot built using Dollyv2, langchain that allows you to chat with your document data conveniently, It works on RAG and embeddings for retrieval.
Summary
A full-stack doctor booking site for both patients and doctors, with specified slots, ML-based clustering used for doctor selection.
Summary
Engineered a Python application, Vid2Pdf, leveraging image processing and subtitle extraction to convert videos into PDF documents, achieving a 95% accuracy rate in text conversion for videos under 10 minutes