Muhammad Ali Ahson

Academic

Verbatim text

~
Muhammad Ali Ahson
+923247748900 I aliahson56@gmail.com I G-11, Islamabad.
Linkedln: httQs:[Lwww.linkedin.comLinLmuhammadaliahsonL
GitHub: htti;1s:Lfgithub.comLMuhammadAliAhson
Education
$achelor of Science (Computer Science)
I
Major:
INLP, ASR, Data Analytics, Data Visualization, Generative Al, Computer Vision
I
Projects
Final Proiect: Kisan Rabta - Agricultural Data Analysis Platform
iTechnologies: WhisperAI, HuggingFace, Tensorflow, Python, Flask, React Native, React.js, Node.js, Google Drive,
PostgreSQL, Google OCR, Supabase
•
Built an Al Voice-to-Data Platform: Developed an automated system with an Al-assistance chatbot,
extracting agricultural data from voice inputs for analysis and user support.
•
Data and Model Optimization: Scraped 18,000 text samples and 2 hours of audio to fine-tune LLM and
Whisper models, achieving 0.001% accuracy in agricultural terms.
•
Multilingual Processing: Converted Urdu transcriptions to English, using NER to detect diseases and issues .
•
Deployment: Launched an Al pipeline on Hugging Face Spaces with APls for seamless access .
•
Applications: Created a recording app and web dashboard with chatbot, offering farmers insights and
recommendations.
Proiects:
Medical Reasoning Chatbot with DeepSeek Rl Distilled LLaMA SB
•
Fine-tuned DeepSeek Rl distilled LLaMA 8B on a medical dataset (Freedomlntelligence/medical-ol-
reasoning-SFT) using Unsloth to develop a reasoning-mode chatbot.
•
Achieved a train loss of 1.4583 over 0.96 epochs with efficient optimization. View Project
Amazon Sales Analysis Dashboard (Power Bl)
•
Built a multi-page dashboard with filters for city, state, shipping methods, and categories .
•
View Dashboard: Amazon Sales Dashboard .
Automatic Number Plate Recognition with YOLOvS
•
Developed a detection model with 97.5% precision, 97.5% recall.
•
Optimized metrics: box loss (0.266), classification loss (0.1776), and distribution focal loss (0.8083) .
Work Experience
Pata Science Intern - ATOMCAMP
July 2024 - Aug 2024
• Built real-time price prediction models for the banking sector to optimize decision-making.
• Leveraged large language models (LLMs) for tools like text generation, blog writing,
$entiment analysis, and image generation.
~I Intern - Datalnsight Lab
Jan 2023 - Aug 2023
• Developed digital solutions for hospitals using YOLO and OCR for patient transcript
digitization, improving model accuracy by 15%.
• Conducted data preprocessing and hyperparameter tuning to optimize model performance.
$kills & Tools
Professional Skills Leadership, Interpersonal Skills, Communication.
jTechnical Skills
Programming Languages: Python, C++, .Net, HTML, CSS, JavaScript, Next.js, React.js
Machine Learning: TensorFlow, Scikit-Learn, OpenCV, YOLO, Pandas, NumPy, Transformers, Lang
:::hain, Dataset, Dialog Chatbot, Botpress, Unsloth.
Web Development: Flask, Django, Hugging face Spaces
Databases: MySQL, PostgreSQL, MongoDB, Supabase
Visualization Tools: Power Bl, matplotlib, Seaborn
~chievements
Dean's List - 2024
rrrainings I Certification
Database Design and Basic SQL in PostgreSQL - University of Michigan
Linear Regression and Modeling - Duke University
nterests
iA.utomation, Data Engineering, Guitar, Football,
FAST NUCES ISLAMABAD CAMPUS

AI enrichment