I'm building Qenot, a study space where college students read, ask,
practice and remember every class in one place.

Ahamed Munsif • Peerlist

<h3>All-in-One Learning Solution</h3><p>Qenot offers a comprehensive study space designed to support your entire learning journey. It consolidates all your learning materials and activities into a single, intuitive platform, eliminating the need for multiple disparate tools. </p><h3>Key Features:</h3><ul><li><p><strong>Collect &amp; Organize:</strong> Gather all your class materials, including lectures, slides, PDFs, YouTube videos, and personal notes, organized by course.</p></li><li><p><strong>Read &amp; Understand:</strong> Highlight key information as you read. Ask questions directly about your documents, and receive answers with citations that link back to the exact page.</p></li><li><p><strong>Practice &amp; Reinforce:</strong> Generate flashcards, quizzes, and written questions from your own study materials. The system provides explanations for incorrect answers to aid understanding.</p></li><li><p><strong>Remember Long-Term:</strong> Utilize spaced repetition (FSRS) to ensure long-term retention. Set daily goals and maintain streaks to stay motivated. Features include a focus timer for dedicated study sessions.</p></li><li><p><strong>Prove Your Knowledge:</strong> Simulate exam conditions with past exams. Get every answer marked and receive a detailed report highlighting areas for improvement. Answers can be submitted via photo, and handwriting is recognized.</p><p> </p></li></ul><h3>Seamless Integration:</h3><p>Qenot replaces the need for separate tools like Quizlet, Notion, Anki, ChatGPT, and PDF readers, providing a unified experience.</p><p> </p><h3>Privacy and Accessibility:</h3><p>Your study material is kept private and secure. Qenot offers a generous <em>Free plan</em> with core features, requiring no credit card to start. Paid plans unlock advanced research capabilities.</p>

Qenot

After months of deep research, engineering, testing, and refinement, our team is beyond proud to officially present AutoGradePro — an AI-powered, production-grade platform that automates and enhances grading for educators. 🔍 What is AutoGradePro? AutoGradePro is an intelligent grading system built to: ✅ Upload and process 200+ student answer scripts in bulk ✅ Automatically extract questions and marking schemes from .pdf, .docx, and .xlsx files and store them in PostgreSQL as structured JSON ✅ Fetch structured marking data to the client for seamless interaction ✅ Use LLMs (Ollama – Qwen2.5:7B) for semantic answer evaluation — not just keyword matching ✅ Grade diverse answer formats: one-word, short phrase, list-based, and numerical-range responses ✅ Deliver fair, AI-assisted scoring for open-ended and semantically varied student answers ✅ Generate instructor-level analytics and performance reports ✅ Support multiple file types: .PDF, .DOCX, .TXT, .XLSX ✅ Fully dockerized, deployed, and performance-tested for concurrent grading requests ⚙️ Tech Stack 🛠 Django | Next.js | Tailwind CSS | Ollama (LLMs) | Docker | AWS (m6i.large) | PostgreSQL 💬 Client Feedback That Meant Everything We were honored to work under the guidance of Dr. Nadeesha Sandamali, Senior Lecturer at the University of Ruhuna. Her mentorship pushed our project far beyond what we thought possible. Here’s an excerpt from her feedback: I personally tested the system using my own question set and over 40 student answer scripts, and found the overall performance of the system to be reliable and effective. The integration of AI for automated evaluation was particularly impressive. The team delivered the project well beyond my expectations with great responsiveness to feedback. I look forward to using this system to streamline the marking of continuous assessments in upcoming modules. We are truly grateful for her trust, wisdom, and continuous support throughout this journey.

AutoGradePro - AI tool for educators to grade

In this project, We designed and implemented a robust Gym Management System Database to handle core operations for a fitness center effectively. The database structure supports functionalities such as member and trainer management, equipment inventory tracking, branch details, and billing and payment records. Here's an overview of the work: 📌 Features: 1. Member Management - Comprehensive data storage for gym members, including personal details, memberships, and assigned trainers. 2. Trainer Records - Tracks trainers' personal details, branch assignments, phone numbers, experience, and salary adjustments using triggers. 3. Branch & Equipment Tracking - Stores branch-specific details (location, capacity, operating hours) and tracks equipment inventory, units, and repair history. 4. Membership Types - Offers various membership plans (Gold, Silver, etc.), with durations and pricing. 5. Schedules & Payments - Manages member schedules and automates payment tracking. 💾 Technical Details: 1. Database Schema Creation - Developed a normalized schema using SQL with 12+ interconnected tables to ensure data consistency and scalability. 2. Relationships and Constraints - Established foreign key relationships for referential integrity. 3. Triggers - Implemented triggers for automated salary adjustments based on trainer experience and to ensure salaries cannot be decreased. 4. Data Integrity - Enforced cascading updates and deletions for maintaining relational integrity across entities. 5. Data Insertion - Populated tables with sample data for branches, members, trainers, and equipment for demonstration. #DatabaseDesign #SQL #GymManagementSystem #Triggers #ERD #UML #BackendDevelopment

Gym Management System Database Design and Imp

Kingdom Domination Game using Pygame & Socket

The Laptop Inventory Management System is a robust software solution designed to efficiently manage the stock of laptops in a store. Utilizing a singly linked list data structure, the system offers functionalities to insert new laptops, update quantities, display stock, and filter laptops based on user preferences such as brand, RAM, and processor. This project showcases advanced C++ programming skills and a deep understanding of data structures. Features: Insertion of Laptops: Allows for the addition of new laptops to the inventory. Merges quantities and updates prices for existing laptop models. Quantity Update: Facilitates the update of laptop quantities in the inventory. Ensures accurate stock management by validating available quantities. Display Stock: Provides a comprehensive display of all laptops in the inventory. Formats the output for easy readability, showing brand, RAM, processor, quantity, and price. Low Stock Alerts: Identifies and displays laptops with low stock levels (below 6 units). Helps in proactive inventory management to avoid stockouts. Preference-Based Filtering: Allows users to filter laptops based on specific preferences such as brand, RAM, and processor. Supports single and multiple preference combinations to refine search results. Technologies Used: C++ Object-Oriented Programming Data Structures (Singly Linked List) Standard Template Library (STL) Project Insights: This project demonstrates the application of object-oriented programming principles in C++ to solve real-world inventory management problems. The use of a singly linked list data structure ensures efficient memory usage and flexible data handling. By integrating functionalities for insertion, updating, and filtering, the system provides a comprehensive solution for managing laptop inventory in a store. #C++ #Data Structures #Singly_Linked_List #Inventory_Management #Object-Oriented #Programming

Laptop Inventory Management System

Built a neural network training pipeline from scratch in C/CUDA and parallelized it six ways to find where distributed training actually stops scaling. Problem: per-residue protein secondary structure prediction (helix/sheet/coil) on the CB513 benchmark — 513 protein domains, 84,119 residues. A 13-residue sliding window, one-hot encoded to 260 dimensions, classified by a 260-256-128-3 MLP (100k parameters). Six implementations, 3,041 lines of C/CUDA, no ML framework: • Serial — hand-written forward pass, backward pass, SGD • OpenMP — data-parallel over the mini-batch, private per-thread gradients • POSIX threads — persistent master/worker pool, two-barrier dispatch • MPI — gradient averaging via MPI_Allreduce, six gradient arrays packed into one flat buffer so each batch costs one collective • Hybrid MPI+OpenMP — two-level decomposition under MPI_THREAD_FUNNELED • CUDA — cuBLAS sgemm, four custom kernels, backward pass fused into the SGD update The design rule that unified all six: keep weights read-only inside the parallel region, give each worker a private gradient buffer, reduce once, apply one update. Result — zero mutexes and zero atomics in the CPU code. Results (i9-14900K 24C/32T, RTX 5090): • CUDA 236.8s → 6.3s end-to-end = 37.8x vs serial • MPI 6.26x per-epoch at 8 ranks (78% parallel efficiency) • OpenMP 5.34x per-epoch at 8 threads • Best CPU layout: hybrid 8 ranks x 2 threads at 0.378 s/epoch, beating both pure MPI and pure OpenMP • All six implementations land within ±0.14 percentage points of the serial baseline — the correctness invariant behind every speedup above Findings: scaling stalls at 8 workers on a 24-core CPU — the fixed ~250 kfloat per-batch reduction overtakes the compute it coordinates, and threads spill from P-cores to E-cores to SMT siblings. Larger batches trade accuracy for throughput (62.3% → 57.0% Q3, batch 64 to 512). Extends the protocol of Zhong et al., J. Supercomputing 41 (2007).

Ahamed Munsif

I'm building Qenot, a study space where college students read, ask, practice and remember every class in one place.