Skip to content
View SobanHM's full-sized avatar

Block or report SobanHM

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
SobanHM/README.md


👋 About Me

I'm a Computer Science graduate from Sukkur IBA University (2022–2026), based in Pakistan, building research-grounded AI systems — Vision-Language Models, RAG pipelines, speech recognition, and computer vision — backed by rigorous benchmarking and evaluation.

  • 🔭 Currently building AI/ML systems spanning VLMs, RAG, and speech AI
  • 📄 Accepted Paper Presenter @ iCOMET 2026Localized Benchmark Evaluation of Vision-Language Models for Visually Impaired Assistance (IEEE Xplore, expected)
  • 💼 Former AI/ML Developer Intern @ Pakistan National Assembly Secretariat
  • 🥇 1st Prize — SIBA Fest 2024 Software Project Exhibition
  • 🌱 Open to AI/ML Engineer, Deep Learning Engineer, and Research Assistant roles

🧠 Skills

Languages
AI/ML & Deep Learning Algorithms
Models & Architectures
AI/ML Frameworks & Libraries
AI/ML Tools & Platforms
Web, Data & Dev Tools

💼 Experience

AI/ML Developer Intern — Pakistan National Assembly Secretariat Jun – Aug 2025 Built a machine-learning speaker recognition system for automated identity verification in a secure institutional environment; handled audio feature extraction, classifier training, evaluation, and confidentiality-aware documentation.

Java Developer Intern — Programmers Tech IT Company Jun – Aug 2023 Built Java/MySQL applications with database-backed modules, CRUD workflows, and API-related processes for business use cases.

Team Lead, Data Management — Alumni & Welfare Club, Sukkur IBA University Aug 2024 – May 2025 Managed member records and reporting workflows, improving data accuracy and cross-team coordination.


📌 Featured Projects

ProjectDescriptionStack
SceneNarrator — VLM Benchmarking & Fine-TuningAssistive scene-narration & navigation system for the visually impaired; LoRA fine-tuned LLaVA-7B, benchmarked against CogVLM/BLIP-2. Accepted @ iCOMET 2026, IEEE Xplore-bound.LLaVAPyTorchPEFT LoRA
Personalized RAG ChatbotDocument ingestion → embeddings → semantic retrieval → grounded generation for knowledge-aware Q&A.PythonRAGChroma DB
Speaker Recognition SystemSingle/multi-speaker recognition pipeline: feature extraction, classifiers, validation & evaluation reporting.PythonAudio Processing
SAM3 vs Human Segmentation ValidationHuman-in-the-loop tool comparing SAM3 masks against ground-truth annotations via IoU, precision, recall, F1.PyTorchSAM3Gradio
Automated Toll Tax / Smart Toll Vision SystemReal-time vehicle detection, tracking & counting for dynamic toll estimation with dashboard overlays.YOLOv8DeepSORT/ByteTrackOpenCV
XAI: Virgin vs Recycled Plastic ClassificationSwin Transformer classifier with Attention Rollout explainability and confidence thresholding.PyTorchSwin Transformer
Fine-Grained Emotion ClassifierCompared classical ML, BERT & RoBERTa on Ekman-emotion data; fine-tuned RoBERTa to 89.0 F1 / 89.4% accuracy.TransformersRoBERTaBERT
Anemia & Leukemia Disease PredictionDual-disease healthcare prediction pipeline with SMOTE balancing and gradient boosting.XGBoostLightGBM
Bank Management SystemSecure desktop banking app with SHA-256 transaction handling and role-based admin controls.JavaMySQL

See GitHub repositories for code.


🎓 Education

BS in Computer Science — Sukkur IBA University 2022 – 2026 Coursework: AI, Machine Learning, Deep Learning, Computer Vision, DBMS, Software Engineering, OOP, Web Development, Cloud & Parallel Computing.


🏆 Publications, Certifications & Awards

  • 📄 iCOMET 2026 — Accepted Paper Presenter, Localized Benchmark Evaluation of Vision-Language Models for Visually Impaired Assistance (IEEE Xplore, expected)
  • 🎖️ Google AI Essentials · Google Prompting Essentials · Google AI Specialization · AI for Research and Insights (badges on Credly)
  • 🥇 1st Prize — SIBA Fest 2024 Software Project Exhibition
  • 🔝 Top-10 — Grand Tech Expo (Bank Management System)
  • 🎗️ Special Mention Award (×2) — Model United Nations, representing the People's Republic of China

📫 Let's Connect



Thanks for stopping by — feel free to explore my repositories!

Pinned Loading

  1. Automated-Toll-Tax-SystemAutomated-Toll-Tax-SystemPublic

    An AI-powered toll collection and vehicle counting system using YOLOv8 and DeepSort. It detects and tracks vehicles, identifies toll-line crossings, calculates fees by type, overlays real-time stat…

    Python 9 6

  2. sam3-human-segmentation-validationsam3-human-segmentation-validationPublic

    A human-in-the-loop segmentation validation system that compares SAM3 automatic masks with human annotations using IoU, Precision, Recall, and F1-score.

    Python 3 1

  3. Spatial-VLM-AssistantSpatial-VLM-AssistantPublic

    A state-of-the-art Assistive Spatial AI system that integrates Vision-Language Models (LLaVA v1.5) with Metric Depth Transformers (Depth Anything V2) to provide real-world spatial awareness.

    Python 1

  4. Anemia_Leukemia_Diseases_PredictionAnemia_Leukemia_Diseases_PredictionPublic

    This repository presents a dual-disease diagnostic system using Gradient Boosting Machines (XGBoost & LightGBM) on clinical blood parameters. The system demonstrates 100% accuracy for both disease …

    Python 4

  5. Benchmark-Evaluation-of-VLMs-for-VIABenchmark-Evaluation-of-VLMs-for-VIAPublic

    Benchmarking Vision-Language Models (VLM) for Visual Impairment Assistance in Pakistan using real-world local environments. This project evaluates VLMs like LLaVA-13B and introduces a human-centere…

    Python 1 1

  6. carbon-visuals-researchcarbon-visuals-researchPublic

    Python 1