Lifely
A Flutter and Firebase app that dynamically prioritizes tasks using mood and energy metrics, with OCR-based note summarization and deadline extraction.
Hi, my name is
NUST graduate building reliable systems with a focus on correctness, performance, and clean user experience. I bridge algorithms, machine learning, and systems engineering.
Hi - I'm Muhammad Athar, a software engineering graduate from NUST with a strong focus on algorithms, machine learning, and building reliable systems.
I started with competitive programming and carried that rigor into research and software projects. That mindset shows up in how I approach correctness, complexity, and edge cases.
Outside code, you'll usually find me solving a Rubik's cube, working through a math problem, or tweaking my Neovim and tmux setup on Arch Linux.
I'm currently open to software engineering roles, research opportunities, and projects that involve strong problem solving.
Terminal-First Workflow
Arch Linux, Neovim, and tmux shape how I work. I like a minimal setup that stays fast and predictable.
Mathematically Minded
Algorithms, proofs, and problem solving are a core part of how I think and build.
Puzzle Solver
Competitive programming, Rubik's cubes, and other structured challenges keep me engaged.
A Flutter and Firebase app that dynamically prioritizes tasks using mood and energy metrics, with OCR-based note summarization and deadline extraction.
A secure AI banking assistant built with a hybrid RAG pipeline, Qdrant, LangChain, and a fine-tuned Qwen2.5-3B model secured with NeMo Guardrails and Presidio.
A hybrid deep learning pipeline integrating SCNet and VGG16 for seizure and pathological waveform detection, with a privacy-preserving LLM report generator using Qwen 3.
A RAG-based therapy assistant built with TinyLlama and FAISS to support personalized, long-term memory across conversations.
A scalable microservices-based video platform on GCP built with Node.js, Express, and MongoDB.
An e-commerce platform for rural women entrepreneurs to manage inventory and orders.
An IoT health wearable prototype with a Flask monitoring dashboard for real-time vitals and GPS location.
A search engine indexing 150,000 NELA-GT 2022 articles with a retrieval pipeline based on token frequencies and barrel partitioning.
A centralized travel management platform built with Django, SQLite3, and Tailwind CSS.
RheinMain University of Applied Sciences
Implemented the Stormer model and optimized global weather forecasting using latent ERA5 data. Devised a flood forecasting pipeline combining ERA5 weather data, river discharge maps, and Sentinel-1 satellite imagery.
National University of Sciences & Technology
Lectured on Git workflows and evaluated quizzes, assignments, and exams for a cohort of 90 students. Curated a HackerRank assignment with 5 algorithmic problems.
TUKL-NUST R&D Center
Worked on classification of EEG signals into normal and abnormal using convolutional and transformer-based models. Proposed a novel architecture for EEG classification that achieved state-of-the-art performance on the NMT dataset, outperforming previous baselines by 10%.
National University of Sciences & Technology
Gold Medalist with a CGPA of 3.91. Relevant interests include data structures, algorithms, operating systems, database systems, and machine learning.
Languages
Libraries & Frameworks
Tools & Platforms
Beyond Code
I'm always open to interesting conversations, collaboration, or new opportunities. Whether you have a question, a project in mind, or just want to talk math and puzzles — feel free to reach out.
Say Hello