Legal-Tech Patents Domain LLM
→
Summary
Fine-tuned an open-source 7B transformer on legal-tech patent filings, optimizing distributed training and data pipelines for patent classification.
Education
Indian Institute of Technology
Jodhpur, Rajasthan, India
→
Bachelor of Science
Applied Artificial Intelligence and Data Science
Courses
Applied Artificial Intelligence
Data Science
Machine Learning
Deep Learning
Reinforcement Learning
About
Highly accomplished Applied AI & Data Science specialist with a Bachelor's from IIT Jodhpur, demonstrating expertise in AI/ML, full-stack development, and MLOps. Proven ability to design, develop, and deploy innovative solutions, evidenced by national competition wins, significant project contributions, and impactful internships at leading technology firms. Eager to leverage strong technical acumen and problem-solving skills to drive advanced AI initiatives.
Projects
Psychology AI Virtual Companion
→
Summary
Developed and deployed a fully open-source, zero-cost AI therapy tool leveraging advanced NLP and MLOps, achieving top user ratings for naturalness and emotional intelligence.
Job Matching System
→
Summary
Developed an AI-powered job matching system using NLP and Machine Learning to match job postings with candidate profiles based on multiple criteria.
Sales Prediction Model
→
Summary
Achieved 3rd place in IIT Goa's StatHack Data Science Hackathon by developing a predictive sales model for VivaRetail, accounting for regional preferences and external factors.
Face Recognition Attendance System
→
Summary
Developed an automated attendance system leveraging facial recognition, real-time video processing, and data management for accurate record-keeping.
Sentiment Analysis in Memes
→
Summary
Developed a multimodal sentiment analysis model for a competition, integrating textual and visual features and addressing class imbalance for enhanced performance.
SMS Spam Classifier
→
Summary
Developed and deployed a machine learning model to classify SMS messages with 98% accuracy, featuring comprehensive data preprocessing and an interactive web application.
Work
Space Marvel AI
|AI Engineer Intern
Remote, Global, N/A
→
Summary
Led the ground-up development of a flagship MCP Client and Server, directly contributing to strategic company scaling and AI research dissemination.
Highlights
Created a proprietary MCP Client and Server from scratch, establishing a core flagship feature for the company's product offering.
Contributed directly to key company decisions, influencing strategic direction and growth initiatives for startup scaling.
Authored technical blogs, disseminating latest advancements in the AI Research field to a broader audience.
Invisible AI
|Operator
Remote, Global, N/A
→
Summary
Conducted critical quality assurance testing for Google DeepMind's Gemini LLMs and optimized data pipelines, enhancing production-grade ML system efficiency and reliability.
Highlights
Performed rigorous quality assurance testing for Google DeepMind's unreleased Gemini LLMs, providing critical feedback to improve model performance and competitive positioning.
Tested the unreleased Gemini model against competitors, contributing to its quality improvement and refinement.
Collaborated with Invisible AI's Operations team to optimize data pipelines and streamline deployment workflows, enhancing the efficiency and reliability of production-grade machine learning systems.
Zetheta Algorithms Pvt. Ltd.
|AI Chatbot Intern
Remote, Global, N/A
→
Summary
Developed a zero-cost educational RAG chatbot, processing 100+ PDF documents into a searchable knowledge base within 8 weeks and reducing inference time by 40%.
Highlights
Developed a zero-cost educational RAG chatbot using Python, FAISS, and Mistral 7B, processing 100+ PDF documents into a searchable knowledge base within 8 weeks.
Implemented a document processing pipeline with PyPDF2 and LangChain, achieving 95% text extraction accuracy from complex educational materials.
Engineered a vector database system using FAISS and all-MiniLM-L6-v2 embeddings, enabling semantic search with 85% retrieval precision.
Integrated a quantized Mistral 7B LLM using llama.cpp for fallback responses, reducing inference time by 40% through CPU optimization.
Designed and deployed a Streamlit web interface with SQLite backend, implementing rate limiting that supported 500+ daily employee queries.
RWS Project Callisto
|Google Search Quality Rater
Remote, Global, N/A
→
Summary
Evaluated and analyzed search results to enhance search engine quality and collaborated on SEO strategies.
Highlights
Evaluated and analyzed search results to enhance search engine quality, relevance, and accuracy worldwide.
Improved search algorithms by providing actionable insights through detailed data analysis, boosting user experience and search performance across global markets.
Collaborated with international teams to optimize search engine optimization (SEO) strategies, contributing to advancements in search technology and algorithm development.
Bluestock Fintech
|Software Development Engineer Intern
Remote, Global, N/A
→
Summary
Led full-stack application development for financial insights across Nifty-100 companies, optimizing system performance through database indexing and caching strategies.
Highlights
Led the development of an innovative full-stack application delivering in-depth financial insights for all Nifty-100 companies, tackling complex data aggregation challenges.
Implemented a high-performance backend using PHP and MySQL, handling large-scale datasets and ensuring data consistency across multiple platforms.
Developed a secure, responsive admin dashboard with Bootstrap 5, facilitating efficient management and updating of extensive financial data spanning over a decade.
Built a dynamic React Native mobile app featuring advanced data visualization techniques, enhancing user engagement with interactive charts and analytics tools.
Optimized system performance through effective database indexing, query optimization, and implementing caching strategies for faster load times.
Outlier AI
|AI Trainer
Remote, Global, N/A
→
Summary
Trained advanced AI models using RLHF for complex Hindi-Latin language pairs, optimizing model accuracy and linguistic nuances through iterative feedback loops.
Highlights
Leveraged Reinforcement Learning from Human Feedback (RLHF) to train AI models on Outlier AI for complex Hindi-Latin language pairs.
Strategically engineered prompts to elicit dual responses, with a focus on identifying and analyzing erroneous outputs for refinement.
Conducted detailed reviews and iterative feedback loops to optimize model accuracy, linguistic nuances, and contextual understanding.