Mariam Lawal

Detail-driven AI Data Annotator and Operations Specialist with hands-on experience producing high-accuracy labelled datasets for machine learning and NLP pipelines. Skilled at reviewing, categorising, and quality-checking diverse data types including text, conversational transcripts, and structured datasets in alignment with strict model-training specifications. Brings a sharp analytical eye, structured thinking, and strong written communication to every annotation workflow.

Experienced working independently and within distributed remote teams to meet throughput targets while maintaining data integrity standards above 97%.

AI Data Annotator & Content Reviewer at Hugo Technologies (2025-02 – Present)

Contract position

• Identified that inconsistent annotation guidelines across a 10-person team were producing conflicting labels; drafted a unified labelling reference document that reduced inter-annotator disagreement by 22% within two weeks of adoption.
• Annotated 18,000+ conversational text samples for intent classification, sentiment polarity, and entity recognition, consistently meeting or exceeding daily throughput benchmarks of 400 items per session.
• Flagged a recurring pattern of mislabelled negation cases that was skewing the model's sentiment scores; worked with the QA lead to create a dedicated edge-case annotation guide that corrected the error class.
• Reviewed and scored AI-generated responses for factual accuracy, tone alignment, and RLHF (Reinforcement Learning from Human Feedback) suitability, contributing directly to model fine-tuning cycles.
• Delivered structured weekly quality reports to the project coordinator highlighting error trends, throughput metrics, and annotation confidence scores.

AI Operations & Data Quality Analyst at DataBridge Solutions (2023-08 – 2024-11)

Full-time position

• Inherited a backlogged dataset of 6,000 unlabelled records with a two-week delivery deadline; reorganised the annotation workflow into category-based batches and cleared the backlog three days ahead of schedule.
• Performed multi-label text classification on customer support transcripts, tagging issues by topic, urgency, and resolution type — output was used to train a routing AI that reduced misrouted tickets by 30%.
• Detected a systematic annotation error in a third-party-submitted dataset where ambiguous prompts were being labelled identically despite different user intents; escalated with documented examples that led to a full batch review.
• Built and maintained a personal annotation tracker in Google Sheets that logged daily output volume, error rates, and category distribution — adopted by the wider team as a standard reporting template.
• Collaborated with content writers to review and rate AI-generated text for coherence, factual grounding, and alignment with brand guidelines across 2,000+ content samples.

Junior Data Labelling Specialist at NexLabel Africa (2023-05 – 2023-07)

Part-time position

• Addressed the challenge of inconsistent category boundaries in a product-classification task by proposing clearer sub-category definitions to the project lead, resulting in a measurable drop in label ambiguity across the team.
• Labelled and validated 9,500+ e-commerce product descriptions for category, attribute tagging, and language quality, maintaining a 98.4% accuracy rate verified through QA spot-checks.

B.A. (Ed) in English Education – Adeyemi Federal University of Education, Ondo (2022)