K Ashritha

About

Computer Science undergraduate specializing in AI/ML and GenAI, with experience building LLM-driven applications, computer vision systems, and backend services. Strong in data structures, algorithms, and problem-solving, with hands-on exposure through hackathons and collaborative projects.

Experience

• Automated Vehicle Damage Detection (Computer Vision / VLM Research Project)
  Company
  Jun 2024 – Oct 2024

Education

• B.E in Computer Science and Engineering
  Keshav Memorial Engineering College
  Computer Science and Engineering · Nov 2022 – Jun 2026

Skills

• Python
• Flask
• PyTorch
• TensorFlow
• RAG Pipelines
• LLMs
• Data Structures and Algorithms
• Oop

Projects

• AI-Powered Investment Decision Agent (Agentic AI System)
  Python, LangChain, FastAPI, Gemini API, Vercel

  Designed a multi-agent AI system using LangChain to simplify stock analysis by combining technical indicators, sentiment, and market trends into a unified decision platform. Integrated Gemini LLM and deployed a full-stack application (FastAPI, Docker, Render, Vercel) enabling real-time BUY/SELL/HOLD recommendations with confidence scores and faster data-driven decision-making.

• Early Diabetes Prediction
  Python, Random Forest, Ollama (Mistral), React

  Developed a machine learning system to predict early diabetes risk from patient health parameters using Random Forest, achieving approximately 96% accuracy for reliable risk assessment. Integrated Ollama using the Mistral model to generate personalized diet and lifestyle recommendations based on symptoms and prediction results.

🏆 Achievements & awards

• Techolution Hackathon — Certificate

  AI-Powered Template Generation Engine. Developed an AI-powered template generation system using RAG pipelines to process PDFs, DOCX, PPT, and images, improving content relevance by approximately 40% and reducing manual editing effort by approximately 50%, achieving response time of approximately 2–4 seconds. Designed a real-time editing interface with instant preview updates and dynamic templates, and built backend services using FastAPI, increasing customization efficiency by approximately 60%.