Maqsudhat Bisiriyu

AI/ML researcher with hands-on experience labeling, cleaning, and quality-checking data for machine learning pipelines, including images, text, and audio. Skilled at following detailed annotation guidelines, working accurately at pace, and flagging edge cases.

Background includes preprocessing and quality control on a multi-country medical imaging dataset, peer review of technical research, and independent evaluation of speech-to-text (ASR) model outputs. Comfortable with remote, deadline-driven, task-based work.

Research Assistant at MIT Critical Data & MoDRIA Initiative (2025-07 – 2026-04)

Multimodal retinal fundus imaging database (Uganda & Kenya)

• Perform data preprocessing, cleaning, and quality control on a large multi-country medical image dataset used for AI model training
• Apply semi-supervised labeling methods to classify retinal fundus images and support secure, structured data pipeline design
• Completed MITx HST.953x coursework in collaborative data science for healthcare (97%)

Independent Project — Audio Transcription & ASR Model Evaluation at Personal Project

• Manually transcribed South African English voice notes and used them to evaluate the output accuracy of OpenAI Whisper against another speech-to-text model
• Compared model transcripts line-by-line against manual ground truth to identify error patterns

Peer Reviewer — AI4Research & AI Bridge MedHealth Workshops at AAAI 2026 (2026)

• Reviewed 13 technical papers against detailed evaluation criteria, providing structured, consistent written feedback

Data Science Researcher at Pioneer Academics / University of Florida (2023-02 – 2023-07)

• Collected, cleaned, and labeled medical imaging data to build breast and cervical cancer detection models (91% accuracy)
• Authored a 27-page research paper documenting data handling and methodology, graded A by faculty reviewer

Software Engineering Intern at eTranzact International PLC (2024-07 – 2024-10)

• Reviewed and labeled transaction data to help build fraud-detection rules, reducing transaction fraud by 30%

B.Sc. in Mathematics – Federal University of Agriculture, Abeokuta (FUNAAB) (2026-08)

non-degree in Cybersecurity & Networking – Prime Innovation Institute of Technology (Cisco & Univ. of Bolton) (2025-04 – 2026-05)