Open to opportunities

Salman Ahmed

@salmanahmed3

Message

AI Engineer specializing in machine learning and computer vision.

Pakistan

Message

What I'm looking for

I am looking for a role that fosters innovation and allows me to leverage my AI skills in impactful projects while promoting a collaborative and growth-oriented environment.

I am an AI Engineer with a strong foundation in machine learning and computer vision, currently leading the development of innovative AI solutions at Taleemabad. My work on the Digital Coach project has significantly improved classroom feedback systems by integrating advanced technologies like Azure Speech-to-Text and GPT-4 Turbo, achieving high compliance with educational standards.

Throughout my career, I have focused on optimizing AI models and enhancing their performance. For instance, I successfully reduced the Word Error Rate of the Whisper model and improved Urdu OCR accuracy using state-of-the-art techniques. My experience also includes building real-time dashboards and architecting pipelines for various AI applications, showcasing my ability to blend technical expertise with practical solutions.

Experience

Work history, roles, and key accomplishments

Current

AI Engineer

Current

Taleemabad

Jan 2025 - Present (5 months)

Led end-to-end development of Digital Coach, an AI-driven classroom feedback system using Azure Speech-to-Text and GPT-4 Turbo, automating observation forms aligned with the World Bank’s Teach Tool. Optimized end-to-end processing to

8 minutes for 30-minute audio via parallel STT/LLM pipelines and GPU acceleration.

Azure Speech To Text GPT 4 Turbo Docker AWS YOLOv11 MLOps Generative AI

AI Intern

AIO

Jun 2024 - Sep 2024 (3 months)

Reduced Whisper model WER from 0.7 to 0.3 using LoRA fine-tuning and CTranslate2, slashing inference time by 65%. Enhanced TTS models (Piper, XTTS) to achieve MOS: 4.2/5 for English synthetic speech.

PIPER Whisper CTranslate2 Piper XTTS Streamlit Machine Learning

AI/ML Research Assistant

Data Insight Lab, NUCES

Aug 2023 - May 2024 (9 months)

Architected a pipeline for digitizing handwritten prescriptions, achieving 97% detection accuracy (YOLOv8) and 96% recognition accuracy (TROCR). Generated synthetic handwriting datasets using RNNs, boosting model robustness by 25%.

YOLOv8 TROCR RNN Machine Learning Computer Vision Data Engineering

Computer Vision and Machine Learning Intern

Data Insight Lab, NUCES

Jun 2023 - Aug 2023 (2 months)

Engineered synthetic datasets to improve entity recognition model accuracy by 18%. This involved applying advanced data generation techniques to enhance model performance.

Machine Learning Computer Vision Data Engineering Data Generation

Education

Degrees, certifications, and relevant coursework

National University of Computer and Emerging Sciences

Bachelor of Science, Computer Science

Studied Computer Science with relevant coursework in Machine Learning, Natural Language Processing, Computer Vision, Data Structures & Algorithms, and Software Engineering. Gained expertise in Cloud Computing (AWS), Generative AI, and MLOps.