Artificial Intelligence undergraduate at the University of Management and Technology, focused on building practical AI systems that move beyond isolated models into complete, usable applications.
Hands-on experience spans speech processing, computer vision, NLP, symbolic reasoning, knowledge graphs, workflow automation, and AI application development. Experienced in Python, TensorFlow, PyTorch, OpenCV, Neo4j, LangChain, Streamlit, and n8n, with practical exposure to data preprocessing, feature engineering, model experimentation, evaluation, inference pipelines, APIs, webhooks, and AI application development.
The goal is to combine AI/ML expertise with software engineering principles to build reliable, practical, and deployable intelligent applications.
- AI Engineering Internships
- Machine Learning Engineering Internships
- AI / ML Engineering Roles
- Applied AI Projects
- Open Source Collaboration
- Research-Oriented AI Projects
| Domain | Proficiency | Details |
|---|---|---|
| Machine Learning | Intermediate | Data preprocessing, feature engineering, model experimentation and evaluation |
| Deep Learning | Intermediate | MLP, CNN, LSTM, GRU and Transformer experimentation |
| Speech AI | Intermediate | Audio preprocessing, MFCC extraction and speech classification |
| NLP | Intermediate | Text processing and contextual AI application development |
| Computer Vision | Intermediate | OpenCV image processing, OCR and document restoration |
| Knowledge Graphs | Intermediate | Neo4j, Cypher, graph construction and multi-hop relationships |
| Symbolic AI | Intermediate | Prolog-based reasoning and hybrid symbolic-AI systems |
| AI Automation | Intermediate | n8n workflows, webhooks and automated processing pipelines |
| AI Application Development | Intermediate | Streamlit interfaces, inference pipelines and end-to-end AI applications |
01 · Acoustic Language Identification
A four-class speech classification system for Urdu, English, Urdu–English code-switched speech, and Arabic using custom and crowdsourced audio recordings.
| Category | Details |
|---|---|
| Stack | TensorFlow · Keras · Librosa · Streamlit · MFCC |
| Scale | 873 audio recordings |
| Performance | Evaluated on 175 unseen test recordings |
| Security | Not a stated project objective |
| Impact | End-to-end speech classification and deployable audio inference |
| Repository | View Repository |
Built an audio preprocessing and feature-engineering pipeline using 16 kHz resampling and 13 MFCC features. MLP, LSTM, and GRU architectures were compared on the classification task.
The selected LSTM model was deployed through a Streamlit interface that accepts audio input and displays the predicted language with per-class confidence scores.
02 · KinGraph AI
A hybrid symbolic-AI system combining natural-language facts, Neo4j knowledge graphs, and Prolog reasoning to derive multi-hop relationships.
| Category | Details |
|---|---|
| Stack | Python · Neo4j · Cypher · Prolog · AIML · Streamlit |
| Scale | Persistent knowledge graph with dynamic facts and relationships |
| Performance | Real-time graph exploration and inference workflow |
| Security | Not a stated project objective |
| Impact | Hybrid symbolic reasoning and graph-based knowledge exploration |
| Repository | View Repository |
The system converts natural-language facts into a persistent Neo4j knowledge graph and uses Prolog-based reasoning to derive multi-hop relationships.
The application includes a reasoning studio and graph explorer displaying dynamic facts, inferred facts, nodes, and graph relationships.
03 · AI Document Text Restoration
An end-to-end document restoration pipeline combining image preprocessing, damage-mask detection, OCR analysis, and contextual LLM-based text reconstruction.
| Category | Details |
|---|---|
| Stack | OpenCV · PaddleOCR · LangChain · Mistral AI · Streamlit |
| Scale | End-to-end document processing pipeline |
| Performance | OCR statistics and reconstruction results presented through the application |
| Security | Not a stated project objective |
| Impact | Automated restoration of damaged document text |
| Repository | View Repository |
Built a document-processing pipeline combining image preprocessing, damage-mask detection, OCR bounding-box analysis, and contextual LLM-based reconstruction.
A Streamlit interface presents original documents, detected damage, OCR statistics, highlighted regions, and reconstructed text.
04 · Agentic Resume Ranking System
An automated workflow for resume intake, text preprocessing, keyword-based scoring, candidate ranking, CSV export, and email delivery.
| Category | Details |
|---|---|
| Stack | Python · n8n · Webhooks · JSON · CSV |
| Scale | Automated multi-stage candidate-processing workflow |
| Performance | Reusable Python ranking logic with local workflow testing |
| Security | Not a stated project objective |
| Impact | Automated candidate ranking and workflow orchestration |
| Repository | View Repository |
Automated resume intake through webhook-driven JSON input, followed by text preprocessing, keyword-based scoring, candidate ranking, CSV export, and email delivery.
Reusable Python ranking logic is separated from the n8n orchestration layer, creating a workflow that can be tested locally and extended.
Mar 2025 – Jun 2025 · Lahore, Pakistan
Contributed to business-development activities surrounding web, mobile, and AI software projects, gaining practical exposure to technology selection, project scoping, quotations, and software-delivery workflows.
Scope of Work
- Researched market demand across web, mobile, and AI software projects.
- Investigated technology-stack requirements and pricing patterns.
- Evaluated client requirements and project needs.
- Gained exposure to project scoping and technology selection.
- Developed practical understanding of quotations and software-delivery workflows.
Skills
Market Research Requirements Analysis Technology Evaluation Project Scoping Software Delivery
| Recognition | Details |
|---|---|
| Academic Standing | CGPA 3.5 / 4.00 in BS Artificial Intelligence |
| AI Engineering | AI Engineer for Data Scientists Associate — DataCamp |
| Graph Technology | Neo4j Certified Professional |
| Graph Data Science | Graph Data Science — Neo4j GraphAcademy |
| Knowledge Graphs | Cypher Fundamentals · Data Modelling Fundamentals · Neo4j Fundamentals |
Build systems that solve real problems — not just models that pass notebooks.