Md. Mehedi Hasan

AI Team, Projsite

• Designed and deployed AI Agent integrated with NoSQL databases to enable real-time, context-aware user interactions.
• Analyzed consumer-level behavioral and interaction data to extract actionable insights and continuously improve AI-driven product functionalities.
• Built and maintained robust MLOps pipelines for automated model retraining, versioning, and continuous delivery to ensure high-availability and performance of deployed models.
• Collaborated with cross-functional remote teams to identify and execute data-driven optimization opportunities.
• Developed scalable, production-grade machine learning infrastructure for seamless integration of AI features across Projsite’s core products and services.

Acote AI Innovation Hub, Acote Group

Adviser: Dr. Mark Kim

• Developed an Object Detection model, successfully resolving significant accuracy challenges faced by the previous model in real-world scenarios.
• Engineered an automated ETL pipeline using Apache Airflow, streamlining data migration from model outputs to the software backend, significantly improving data accessibility and operational efficiency.
• Developed a Temporal Tracking-based automated data collection platform, leveraging CCTV footage to reduce manual effort and optimize data collection.

Team Helios, AlterSense Limited

Adviser: Dr. Nabeel Mohammed

• Architected a real-time ML pipeline to ingest high-throughput camera streams (1.1 GB/s) and stream data into Apache Kafka for scalable processing.
• Developed core data ingestion modules in C++, utilizing advanced concurrency techniques (mutexes, condition variables, thread pools) for thread-safe, high-performance streaming.
• Designed and deployed a data warehouse solution to store both raw and processed data, facilitating real-time analytics and scheduled batch inference.
• Implemented scheduled inference workflows that retrieved data from the warehouse for deeper analysis, ensuring temporal consistency and model robustness.
• Built a horizontally and vertically scalable distributed system, ensuring high availability, fault tolerance, and consistent performance under sustained load.
• Prioritized architectural principles of scalability, maintainability, and reliability to support ongoing high-load operations and ensure seamless deployments.

Team Helios, AlterSense Limited

Adviser: Dr. Nabeel Mohammed

• Optimized multiple surveillance vision models, reducing GPU memory overhead by 10× through performance profiling with NVIDIA Nsight Compute.
• Delivered a robust object detection model in noisy environments, achieving an F1 Score of 0.722 by effectively addressing data imbalance challenges.
• Employed TensorRT framework to decrease GPU VRAM usage by 30 % and inference speed by 1.7×, enabling deployment on low-spec hardware.
• Developed and deployed an end-to-end real-time vision inference pipeline for local servers, utilizing distributed computing platforms for enhanced performance.
• Engineered algorithms to automate a couple of manual software workflows on deployed machine learning models, boosting performance and scalability by 25 %.
• Provided mentorship and strategic guidance to a junior team member, leading to the successful development of a new feature for an existing product.

Department of Electrical and Computer Engineering, North South University

Supervised by: Dr. Mohammad Rashedur Rahman, Mirza Mohammad Lutfe Elahi, and Silvia Ahmed

• Delivered a high-performance website for the ICCIT 2021 Conference, exceeding 30,000+ traffic, and received positive feedback from attendees.
• Conducted a livelihood vulnerability assessment solution for 10,000+ people for post-disaster environment, enabling targeted relief distribution.
• Migrated a legacy PHP project to Django, driving a 20% performance gain and unlocking future scalability.
• Developed and deployed an Android app for precise data collection, reducing errors by 10% and optimizing field operations.

Department of Electrical and Computer Engineering, North South University

Worked and collaborated with Dr. Nabeel Mohammed, Dr. Sarker Tanveer Ahmed Rumee, and Sarnali Basak

• Conducted tutorial sessions for students needing extra help outside of class hours.
• Graded home-works and assignments.
• Stayed informed about test dates, times, and other course-related deadlines.
• Maintained 04 hours per week per section divided among the assisting faculty members for student consultation.
• Assisted faculty members in their course-related work except for grading quiz/exam papers.

Department of Electrical and Computer Engineering, North South University

• CGPA: 3.77 / 4.00 (90-92% marks)
• Graduated with Magna Cum Laude.
• Thesis Title: Inter-Dataset Critical Evaluation of Common Object Detection Model.
• Achieved 2nd Runner Up position in Electrathon 2018 organized by IEEE NSU.