Hi there!
I’m Kishan Yadav

I'm into  

AI Engineer — Full-Stack AI & LLM Systems

RoboAIAPaths, Faridabad
Sept 2025 – Present

• Architected and developed ASB, a production-grade AI numerology platform with a modular FastAPI backend and a branded Streamlit interface.
• Engineered the complete numerology computation engine including core calculations, yearly/monthly/daily predictions, profession traits, health diagnostics, and rule-based special-number logic.
• Built a provider-agnostic LLM subsystem (OpenAI, Ollama, Mock) with strict JSON schemas, deterministic prompts, and resilient fallback heuristics.
• Implemented SaaS-ready backend features: secure API gateway, feature gating, caching layers, environment-based configuration, and structured error handling.
• Developed automated PDF pipelines using ReportLab and created custom matplotlib visualizations for unified AI + numeric advisory reports.
• Collaborated on UI/UX, product flow, and branding to deliver a polished and reliable AI-driven numerology advisory platform.

Data Science Intern → Junior Data Scientist

Global Source Technology, Nashik
Sept 2024 – Aug 2025

• Performed EDA, preprocessing, and data cleaning on structured datasets as part of real-world analytics initiatives.
• Developed and evaluated supervised ML models including Logistic Regression, Random Forest, and XGBoost using Scikit-learn.
• Created clear visual reports and trend analyses using Matplotlib and Seaborn for internal presentations and insights.
• Collaborated with senior data scientists on feature engineering workflows and ML pipeline refinement.
• Gained hands-on exposure to end-to-end ML pipelines, documentation standards, and Git-based version control in an on-site environment.
• Simultaneously completed a 6-month Full Stack Data Science & AI program (Jan–Jun 2025), strengthening applied ML and industry practices.

ASB: AI Numerology Platform

A production-grade AI numerology platform featuring a custom computation engine, provider-agnostic LLM subsystem (OpenAI, Ollama, Mock), strict JSON-schema workflows, deterministic prompt flows, and resilient fallback logic. Includes automated PDF reporting, custom visualizations, and a modular SaaS-ready FastAPI backend with a branded Streamlit frontend.

Portfolio RAG Chatbot

An AI chatbot built with embeddings + retrieval that answers questions about my projects, experience, and skills. Uses LangChain, FAISS, and local LLMs for fast and contextual responses.

MockMate: Virtual Mock Interview Assistant

MockMate is an AI-powered virtual mock interview platform that generates personalized questions from uploaded resumes and supports voice answers with real-time feedback using Ollama + Mistral-7B.

StyloGenie: AI Wardrobe Assistant

An AI-powered outfit recommendation app using LangChain, Streamlit, and color theory. Suggests clothing based on user input and displays visual matches via Pexels API.

EcoVision: Intelligent Trash Sorting

Real-time garbage classification system using MobileNetV2 and OpenCV to detect plastic, glass, metal, and organic waste for smart bins with >90% accuracy.

DreamSpaceAI: Redesign Your Room

Transforms room photos into styled designs using Stable Diffusion XL and T2I Adapter with support for voice prompts. Ideal for interior designers and architects.

Text-to-Image with Stable Diffusion

Text prompt to image generation using Hugging Face + SD v2.1. Explores GenAI deployment with prompt engineering and GPU acceleration.

JobGuard: Fake Job Posting Detection

A Streamlit-based ML tool that flags fake job listings using TF-IDF and RandomForest. Empowers job seekers with smart prediction and a clean UI.

AI-Powered EDA App

Automated EDA tool using Pandas, Seaborn, Matplotlib, and Mistral-7B. Includes Gradio UI for CSV uploads and real-time AI-powered insights.

Sales Data Analysis Web App

Interactive dashboard for analyzing rental store sales using Streamlit, Matplotlib & Seaborn. Ideal for retail analysts and e-commerce insights.

Avocado Price Forecasting

Regression model comparison for forecasting avocado prices. Includes visualized insights using real retail data from 2013–2018.