Aditya Ranjit Kotwal

Software Engineering Intern @ weav.ai

Website is still a WIP, all GitHub links will be made available shortly

🌟🌟 Shaprnr

Built an Agentic-AI powered academic assistant: Sharpnr - designed to eliminate student overwhelm by centralizing, prioritizing, and personalizing academic workflows.

Architecture

Features

  • Real-time academic update aggregation via Canvas, Slack, Email, and Google Calendar.
  • Personalized study support using V.A.R.K (visual, auditory, read/write, and kinesthetic) based AI learning models.
  • Fully on-premise privacy-first deployment using FastAPI, Ollama, MongoDB, and Apache Airflow.
  • Smart reminders, lecture summarization, and adaptive quiz prep powered by multi-agent AI orchestration.

Achievements

  • Achieved 100% local data privacy with no cloud dependency; ensuring complete student data ownership.
  • Boosted study efficiency and task prioritization with a multi-agent, multi-model AI system (Mistral, Qwen).
  • Delivered an intuitive student experience by blending natural language planning agents with real-time task orchestration.
  • Empowered personalized learning by dynamically adapting V.A.R.K scores based on performance data and quiz feedback.

Technologies Used

Python

Python

FastAPI

FastAPI

React

React

Mistral

Mistral

MongoDB

MongoDB

NextJS

NextJS

Ollama

Ollama

Slack API

Slack API

Tailwind CSS

Tailwind CSS

Google Calendar

Google Calendar

Docker

Docker

Apache Airflow

Apache Airflow

🌟🌟 MindGraph

Built an Agentic-AI powered educational assessment platform: MindGraph – designed to revolutionize academic testing by leveraging AI-driven content analysis, automated question generation, and intelligent student engagement systems.

Architecture

Features

  • AI-powered content analysis via OpenAI Whisper and Faster-Whisper for audio transcription and document processing
  • Intelligent question generation using LangChain and Google Gemini models to create adaptive quizzes from educational content
  • Dual user ecosystem with professor management and student enrollment systems for comprehensive academic workflow management
  • Modern glassmorphic UI built with Next.js 15, React 19, and Tailwind CSS for an intuitive educational experience
  • Real-time audio processing with support for PDF, DOCX, and TXT uploads for automated content extraction
  • Adaptive learning system featuring agentic tutoring with retry logic and personalized feedback mechanisms

Achievements

  • Implemented comprehensive dual-user system with separate professor and student portals, enabling complete academic workflow management
  • Developed intelligent content processing pipeline that auto-generates contextual questions from uploaded documents and audio files using advanced AI models
  • Created adaptive learning algorithms with agentic tutoring that delivers personalized feedback, retry logic, and mastery tracking
  • Built modern, accessible UI with glassmorphic design for a smooth, intuitive user experience
  • Established robust backend architecture with ObjectId serialization, global middleware, and production-ready error handling
  • Integrated real-time audio transcription, allowing professors to generate assessments directly from lectures and spoken content

Technologies Used

Python

Python

FastAPI

FastAPI

NextJS

NextJS

MongoDB

MongoDB

Tailwind CSS

Tailwind CSS

Docker

Docker

Apache Airflow

Apache Airflow

AWS S3

AWS S3

AWS Lambda

AWS Lambda

AWS ECS

AWS ECS

Neo4j

Neo4j

Gemini

Gemini

KierAlign

Built a visually rich, interactive learning tool: KierAlign – designed to teach the Needleman-Wunsch algorithm through immersive, animated matrix visualizations inspired by the Lumon Terminal Pro interface from Apple TV+’s Severance.

KierAlign Demo

Features

  • Dynamic matrix animation showing how alignment scores are computed in real time
  • Interactive traceback path visualization to highlight optimal sequence alignments
  • Multiple alignment discovery when alternative solutions exist
  • Aesthetic, terminal-inspired interface that mimics the Lumon Terminal Pro
  • No setup required – just open index.html in any browser to start learning

Achievements

  • Enabled hands-on learning of the Needleman-Wunsch algorithm for bioinformatics education
  • Created a fully client-side experience with zero server dependency
  • Built a lightweight, responsive layout that supports interactive step-by-step learning
  • Delivered a retro-futuristic UI design experience using IBM Plex Mono and custom color themes
  • Visualized real-time algorithm decisions to reinforce student understanding of global sequence alignment

Technologies Used

JavaScript

JavaScript

HTML5

HTML5

CSS3

CSS3

D3.js

D3.js

🌟 Scalable Face Recognition as a Service

Engineered a high-performance face recognition service leveraging AWS cloud services to process concurrent video streams efficiently.

Pipeline

Features

  • Real-time face recognition with Lambda, S3, SQS, and EC2.
  • Custom auto-scaling algorithm for optimizing resource allocation based on real-time metrics.

Achievements

  • Improved processing throughput by 3x with a custom auto-scaling algorithm.
  • Achieved efficient and scalable performance for high-concurrency video analysis tasks.

Technologies Used

Python

Python

FastAPI

FastAPI

AWS EC2

AWS EC2

AWS ECS

AWS ECS

AWS Lambda

AWS Lambda

AWS S3

AWS S3

AWS SQS

AWS SQS

🌟 Go-Based Redis Clone

Engineered a lightweight in-memory database system in Go, inspired by Redis, focusing on high-performance data handling and fault tolerance.

Go-Based Redis Clone

Technologies Used

Go

Go

Redis

Redis

🌟 Scalable Banking Ecosystem

Developed a robust and scalable banking application leveraging FastAPI to implement core principles of distributed systems. The project focused on enhancing system reliability and performance for high-concurrency environments.

Banking System gRPC

Key Features

  • Distributed Systems Implementation:
    • Incorporated remote procedure calls, logical clocks, and client-centric consistency models to ensure seamless operation across distributed nodes.
  • Client-Centric Consistency:
    • Ensured high data accuracy and integrity across multiple nodes by implementing client-centric consistency techniques.
    • Maintained robust performance under a load of 100+ concurrent requests.

Achievements

  • Designed and implemented a scalable architecture supporting distributed system principles.
  • Optimized the application to handle high-concurrency scenarios efficiently.
  • Ensured data integrity and accuracy across distributed nodes, even in the presence of system faults or network delays.

Technologies Used

Python

Python

FastAPI

FastAPI

gRPC

gRPC

Grammar Play

A C++ program to analyze context-free grammars. The program reads a grammar description and performs specific tasks based on command-line arguments. These tasks include printing lists of terminals and non-terminals, calculating FIRST and FOLLOW sets, left factoring the grammar, and eliminating left recursion.

Sample Input:

S -> C B C D A B C *
A -> C B C D *
A -> C B C B *
A -> C B D *
A -> C B B *
B -> b *
C -> c *
D -> d *
#

Technologies Used

C++

C++

Advanced Tic-Tac-Toe Mobile Game with AI Opponent

Developed a mobile game with advanced AI, designed to provide a challenging and engaging experience for users of varying skill levels.

Technologies Used

Kotlin

Kotlin

SQLite

SQLite