Ayush Pradhan Email & Phone Number

Mumbai, Mh, In

India

1. Designed and deployed an Amazon Seller Bot leveraging Groq framework and Llama3.1-70b, enabling ASIN-based product retrieval and file-based querying (CSV, Excel, Images).2. Optimized RAG workflows by implementing advanced preprocessing with NLTK (stopword removal, tokenization), enhancing data relevance from S3 bucket retrieval.3. Integrated OpenAI Assistant API for file-based queries, refining prompt tuning for scope-specific, product-focused responses, increasing query accuracy.4. Developed "Agents" productivity tool, streamlining multi-prompt workflows with Google Search integration, reducing task completion time by 95% and raising user satisfaction by 20%.5. Streamlined API deployment using FastAPI and Docker, reducing deployment time by 40% while scaling efficiently to meet user demand.

Mumbai, Maharashtra, India

1. Designed and implemented a Brand Sustainability Rating Tool, automating brand analysis through advanced data pipelines, NLTK-powered data cleaning, and chunking for Vector DB storage.2. Improved RAG similarity search with metadata filtering, enabling brand-specific queries, resolving rate-limiting challenges, and accelerating actionable reporting by 85%.3. Developed an EU Guidelines Checker using GPT-4o, LangChain, and regulatory frameworks, reducing compliance violations by 25%.4. Streamlined AI deployment workflows with automated Docker-based containerization, enhancing performance and reducing operational overhead by 30%.

Mumbai, Maharashtra, India

1. Developed an AI-powered QA bot to parse PDF documents, store content in a vector database, and facilitate context-aware question-answering, reducing manual lookup time by 60%.2. Engineered an AI Diet Planner that calculates BMI from user inputs and delivers tailored Indian diet plans, enhancing recommendation accuracy by 45% and boosting user satisfaction by 30%.

Thane, Maharashtra, India

As an intern, I worked on two machine learning projects, one aimed at predicting the yearly amount spent by users on an e-commerce platform using features such as time spent on the application and website, length of membership, and others, and the other aimed at predicting house prices in the USA using features such as area income, number of rooms, and age of house.In both projects, I used machine learning techniques such as linear regression to build predictive models that could accurately estimate the target variable based on the given features, thus providing valuable insights for businesses and individuals looking to make informed decisions.

Thane, Maharashtra, India

Designed an enhanced version of existing Student Management System with Python 3.8 technology, leading in a considerable increase in student performance Visualization and Record management.This contribution boosted the project's efficiency by 30% over its prior iteration, allowing the school to have a better understanding of its kids' academic development.