I am Faizan Riasat, a dedicated Software Engineer with over two years of experience in AI, machine learning, and full-stack development. My passion lies in designing scalable AI/ML pipelines, optimizing OCR systems, and deploying robust MLOps frameworks. Currently, I am pursuing my Master's in Applied Computer Science at Georg-August-Universität Göttingen, Germany, where I continually seek to expand my knowledge and skills in cutting-edge technologies. I thrive on tackling complex challenges and delivering innovative solutions that make a difference.

• Email: riasatfaizan468@gmail.com
• LinkedIn: Faizan Riasat

Stealth Startup, Heidelberg, Germany
Aug 2023 – May 2024

• Developed and implemented a Tesseract OCR system to extract text from PDFs and images, integrating a large language model (LLM) to convert unstructured text into structured JSON for downstream applications.
• Established evaluation metrics, including Average Embedding Score and Extrema Score, utilizing GoogleNews-vectors-negative300 to gauge the effectiveness of chatbot responses.
• Conducted thorough research and comparative analyses of various OCR models from Hugging Face, Google Cloud Vision, and Microsoft Azure, enhancing model selection for improved accuracy and performance.
• Collaborated in the design and execution of a comprehensive mocking system for end-to-end (E2E) testing, optimizing the testing workflow and boosting development efficiency.
• Engaged actively in Agile processes, participating in bi-weekly standups, retrospectives, and sprint planning sessions to ensure project milestones were achieved and features were delivered promptly.

Wateen Telecom, Remote
Aug 2022 – June 2023

• Engineered a robust ETL pipeline utilizing BigQuery to analyze sales data, facilitating predictive analytics and trend identification, which enhanced decision-making and forecasting accuracy.
• Implemented a real-time tracking system for shopping mall visitors, utilizing gender detection and age estimation models to derive valuable insights into visitor demographics and foot traffic behavior.
• Designed and launched a face verification system using AdaFace, achieving 99% accuracy on Streamlit to automate employee attendance management.
• Developed a review classification system that categorized customer feedback, resulting in a 10% increase in customer satisfaction through improvements in product and service quality.
• Automated the extraction and storage of electricity bill data using Selenium, completely eliminating manual data entry and reducing processing time by 100%.

Orel Vision, Remote
Apr 2021 – Feb 2022

• Trained and optimized a Masked R-CNN model for precise detection of tree trunks, calculating essential forestry metrics such as diameter at breast height, above-ground biomass, and carbon content with only a 2% margin of error, supporting environmental sustainability initiatives.
• Developed and deployed a YOLO model on Jetson Nano for real-time detection of various crop types, enhancing agricultural monitoring and enabling prompt assessment of crop health and growth.
• Built and implemented a time-series forecasting model using Prophet for crop yield predictions, aiding farmers in effective resource management. ion.

• Developed a complete sales forecasting system using time-series analysis, integrating models such as ARIMA, Prophet, and LSTM to predict sales trends and seasonality.
• Deployed a real-time dashboard for forecast visualization, enabling inventory optimization and strategic planning.
• Tools Used: Python, AWS, Flask, ARIMA, Prophet, LSTM

• Built an NLP-based system for analyzing customer feedback across channels (social media, emails, and surveys), leveraging BERT for sentiment analysis and topic modeling.
• Deployed a cloud-based REST API for real-time feedback classification and provided a dashboard for sentiment trends and actionable insights.
• Tools Used: Python, BERT, Django, REST API, AWS

• Developed a computer vision system for real-time inventory monitoring using IoT-enabled cameras and YOLOv5, with automated alerts for low stock.
• Implemented a cloud-based dashboard for tracking inventory, predicting shortages, and optimizing reorder times with SKU-level insights.
• Tools Used: Python, YOLOv5, OpenCV, Flask, AWS IoT, Streamlit

• Created a machine learning pipeline to detect fraudulent transactions using supervised learning and anomaly detection algorithms.
• Deployed the solution via Docker and Kubernetes for scalable, real-time fraud detection, with a custom-built analytics dashboard to track fraud patterns.
• Tools Used: Python, XGBoost, Random Forest, Isolation Forest, Docker, Kubernetes, AWS