Software Engineer with 2+ years in AI/ML & Cloud Technologies
AI is analyzing your overall score…
Identifying your key strengths…
Evaluating your skill match against the job requirements…
Assessing your cultural and operational fit
Highly motivated Software Engineer pursuing a Bachelor of Science in Computer Science with a concentration in Artificial Intelligence. Possessing 2.5 years of experience across various roles including Research Assistant, Software Engineering Intern, and Machine Learning Intern. Proficient in Python, AI/ML frameworks (PyTorch, TensorFlow), cloud platforms (AWS), and diverse database and backend technologies. Proven ability to develop deep learning pipelines, implement AI/ML workflows, and build scalable data services, with a strong focus on innovative solutions and problem-solving.
FAST, National University of Computer and Emerging Sciences
Bachelor of Science · Computer Science (Concentration: Artificial Intelligence)
August 1, 2022 – June 30, 2026
Genesys Research Lab, FAST NUCES
Research Assistant (Part-Time)
September 1, 2025 – June 1, 2026
Islamabad, Islamabad Capital Territory, Pakistan
SKAI Worldwide
Remote Software Engineering Intern
April 1, 2025 – October 1, 2025
Seoul, Seoul, South Korea
FAST NUCES
Teaching Assistant & Lab Demonstrator
August 1, 2024 – May 1, 2025
Islamabad, Islamabad Capital Territory, Pakistan
Genesys Research Lab, FAST NUCES
Machine Learning Intern
June 1, 2024 – September 1, 2024
Islamabad, Islamabad Capital Territory, Pakistan
Financial Sentiment Analysis & Topic Modeling
June 1, 2026 – Present
Engineered an experimental NLP pipeline on financial text incorporating LDA topic modeling to cluster semantic themes. Evaluated FinBERT, zero-shot/few-shot local LLMs, and a custom FAISS-based RAG architecture using sentence embeddings to compare retrieval-enhanced reasoning and classification accuracy.
Image Denoising & Variational Autoencoders
June 1, 2026 – Present
Designed a convolutional Denoising Autoencoder to reconstruct CIFAR-10 images from injected Gaussian and salt-and-pepper noise. Implemented a VAE on Fashion-MNIST using the reparameterization trick to sample latent vectors for synthetic image generation, and evaluated reconstruction and KL-divergence losses.
DevFlow – AI Requirements Assistant
June 1, 2026 – Present
Built an interactive requirements assistant with Python and LangGraph using Gemini LLMs, implementing state management with Pydantic and LangGraph's StateGraph for multi-step interactions. Automated DOCX requirements document generation with python-docx.
Vision Transformer (ViT) vs. CNN Architectures
June 1, 2026 – Present
Implemented a custom Vision Transformer featuring image patching, linear embeddings, positional encoding, and multi-head self-attention to evaluate attention-based modeling against standard CNN architectures (with batch normalization and dropout) on the CIFAR-10 dataset. Designed fully-connected ANN models and hybrid CNN+ANN feature extractors from scratch without pre-trained weights.
Neural Machine Translation (English-to-Urdu)
June 1, 2026 – Present
Built an NLP translation system using a vanilla RNN encoder-decoder architecture, handling word-level tokenization and sequence encoding with padding masks. Performed systematic grid search hyperparameter tuning and implemented greedy and beam search inference decoding.
Real-Time Crime Analytics Pipeline
June 1, 2026 – Present
Built a real-time streaming data pipeline using Kafka and Storm, alongside PySpark for batch analytics and ETL formatting of time-series CSV data. Applied K-Means clustering to location data to identify crime hotspots, exporting metrics to PostgreSQL and MongoDB for a Streamlit visualization dashboard.
MedBot - AI Medical Assistant
June 1, 2026 – Present
Developed an AI-driven medical assistant, initially built using a quantized Llama 3.1 8B model trained on a medical dataset, then transitioned to the Google Gemini 1.5 Flash API due to hardware limitations. Built the backend with Python, Flask, TensorFlow, and PyTorch, with a frontend in HTML, CSS, and JavaScript.
TennisVision - Final Year Project
June 1, 2026 – Present
Built an end-to-end deep learning pipeline to analyze single-camera tennis footage for player biomechanics benchmarking against professional references, covering player detection, pose estimation, ball tracking, hit event localization, and stroke classification. Employed YOLOv8 (fine-tuned for tennis court player detection), RTMPose-Large for 2D pose estimation with SmoothNet for temporal smoothing, and TrackNet with an inpainting network for ball trajectory tracking and occlusion recovery. Adapted F3ED (a feature extractor + GRU + MLP architecture from the F3Set benchmark) for tennis hit event detection from pose and ball trajectory features. Fine-tuned the Badminton Stroke-type Transformer (BST) – originally trained for badminton – on the F3Set tennis split (7 stroke classes) to perform stroke classification; addressed class imbalance of underrepresented strokes (volley, drop) via keypoint mirroring, Gaussian noise injection, and speed perturbation augmentations, improving accuracy from 88% on the raw imbalanced dataset to 93%. Achieved a per-shot inference time of ~9 seconds on an RTX 2060 across the full pipeline. Served the pipeline through a FastAPI and Celery backend with Redis and MongoDB, paired with a React frontend for real-time dashboards and synchronized video playback.
Predictive Modeling of California Housing Prices
June 1, 2026 – Present
Conducted exploratory data analysis and benchmarked Linear Regression, SGD, and Polynomial models using K-fold cross-validation. Natively implemented batch and stochastic gradient descent optimization algorithms in NumPy.
Multi-threaded Pac-Man
June 1, 2026 – Present
Built a multi-threaded game in C++ and SFML, implementing synchronization techniques using mutexes, semaphores, and Pthreads to manage UI, Pac-Man, and ghost AI on separate threads to avoid deadlocks.
Online Event Booking Platform
June 1, 2026 – Present
Designed a scalable, polyglot microservices architecture using Python (Flask, FastAPI), Node.js, and Spring Boot with PostgreSQL and MongoDB. Automated deployment on an AWS Kubernetes cluster using a CI/CD pipeline with Docker, Terraform, and GitHub Actions.
Simulation of IPFS (Decentralized Storage)
June 1, 2026 – Present
Simulated the InterPlanetary File System (IPFS) in C++ to study peer-to-peer distributed file storage, using a Distributed Hash Table (DHT) for routing and a B-Tree for storage optimization.
Advanced Learning Algorithms
Stanford/Coursera
June 1, 2026 – Present
Git & GitHub
IBM
June 1, 2026 – Present
Supervised Machine Learning
Stanford/Coursera
June 1, 2026 – Present
Introduction to Cloud Computing
IBM
June 1, 2026 – Present
Developing Back-End Apps with Node.js
IBM
June 1, 2026 – Present
Cultural Fit Analysis
The candidate's diverse academic projects, involvement in research labs, and interest in various technical domains (AI, hardware-software integration, automotive mechanics) suggest a curious, driven, and adaptable individual. Their participation in programming competitions indicates a competitive spirit and a willingness to tackle challenging problems, which aligns well with a culture of innovation and continuous improvement. The academic focus, however, means their experience with corporate cultural norms and team dynamics is less evident.
Soft Skills & Operational Fit
The candidate's project descriptions indicate strong problem-solving skills and an ability to work on complex, multi-faceted technical challenges. Their role as a Teaching Assistant suggests good communication and mentoring abilities. The academic nature of most projects, however, means direct operational fit in a fast-paced commercial environment needs further validation.