Jotesh Kumar

Profile Picture 6a0fe60ce444a

Jotesh Kumar

Graduate of 2026
BS Computer Science

Aspiration Statement

CS graduate driven by AI-powered products, full-stack engineering, marketing, and business development. I build things that work, from creating apps to pop-up festivals, and intend to keep doing both.

Core Skills

  • C++
  • Figma
  • Marketing
  • Node.js/React
  • Python

Core Competencies

  • Acts with Ownership
  • Adaptability
  • Collaborates Openly
  • Planning
  • Problem Solving
  • Takes Initiative

Preferred Career Paths

First priority: Marketing Manager

Second priority: Software Engineer

Third priority: Business Development Manager

Experience

Leadership / Meta-curricular

  • Deputy Socials Director
  • Sponsorship Lead
  • Member, Computer Science And Engineering Club
  • Member, Feminist Collective
  • Member, Young Leaders Club

Internship / Volunteer Work

  • Content Creator, Lime Studio (September 2025 – November 2026)
  • Manager & Partner, Local Popup Karachi (December 2024 – April 2026)
  • Marketing Executive & Content Creator, Hr Consultants (February 2025 – April 2026)
  • Business Development and Marketing Intern, Techexons (June – August 2024)

Publications / Creative Projects

  • Exchange Program – Got into UC Berkely for Habib University's Learn Abroad Program

Final Year Project

Project Title

Ilm Dost: An AI-Powered Socratic Math Tutoring System for Primary School Students

Description

Ilm Dost is a full-stack web application designed to improve math comprehension among underprivileged Grade 5 students aligned with the Sindh Board curriculum. The system pairs a React/TypeScript frontend with a Node.js/Express backend and PostgreSQL database, deploying a fine-tuned Phi-4 language model via LoRA adapters to deliver Socratic tutoring, guiding students through reasoning rather than providing direct answers. The platform supports student and guardian roles, OTP-based login, quizzes across seven curriculum chapters, voice input via the Web Speech API, and engagement mechanics including streaks and badges. A curated evaluation dataset benchmarks the fine-tuned model against the base model using dual rubrics assessing pedagogical quality and mathematical accuracy. The project aims to address Pakistan's learning crisis through accessible, conversational AI tutoring.

Project Pictures