Research Assistant(Data Science and Analytics)

Key Responsibilities

1. Design, build and maintain end‑to‑end AI data pipelines for construction‑site AI projects, including multi‑modal (RGB + IR) dataset acquisition, pre‑processing, alignment, quality control via clustering methods (KMeans, HDBSCAN), data augmentation and annotation standardisation; improve dataset consistency for facade defect detection, on‑site hazard detection and intelligent tower crane modules.
2. Develop, train, tune and benchmark deep‑learning models using PyTorch for construction computer‑vision tasks: object detection (YOLO family), keypoint detection and pose estimation for tower‑crane hook stabilisation and site safety monitoring. Measure recall / precision / accuracy and iterate augmentation strategies and model configurations to lift real‑world performance metrics.
3. Support data engineering for the AI‑powered Intelligent Tower Crane System: manage on‑site raw data collection, cleaning and validation for hazard detection and hook‑stabilisation modules; analyse system performance and contribute to accuracy and operational efficiency improvements.
4. Implement Retrieval‑Augmented Generation (RAG) pipelines for internal research; integrate multi‑retriever modules (BM25, dense retrievers, ColBERT, E5 / GTE‑Qwen‑Instruct), open‑source LLMs, conduct quantitative evaluation using MAP, NDCG, precision and recall metrics. Explore potential LLM / agent‑AI use cases for construction‑site documentation, incident analysis and technical knowledge retrieval.
5. Build NLP solutions such as text tokenisation, feature extraction and supervised text classification; process structured / unstructured text data including project reports, site incident records and technical documentation.
6. Perform statistical analysis and exploratory data analysis (EDA) using Python and R; conduct quantitative analysis on construction‑site operational data, safety incident datasets and experimental model outputs.
7. Construct and optimise ETL workflows; manage SQL and NoSQL (MongoDB) databases, integrate multi‑source datasets to build AI‑ready data infrastructure for construction‑tech research and decision‑making.
8. Collaborate cross‑functionally with research and engineering teams, analyse model error cases from real‑world site data, document experimental workflows, model performance and technical findings; support report generation and data visualisation via Power BI / Matplotlib / Seaborn.
9. Keep track of state‑of‑the‑art advances in LLM, agentic AI systems, multimodal deep learning and computer vision; prototype new techniques and adapt them for construction robotics, site safety and intelligent tower‑crane research projects.

RequirementsEducation

• Master’s degree in Data Science & Analytics, Statistics, Computer Science, or related quantitative disciplines; Bachelor’s degree in Statistics or relevant STEM field is required.
• Exposure to multimodal learning, computer vision or applied data analytics for engineering domains is an advantage.

Technical Competencies

• Strong programming skills in Python, PyTorch; practical experience with R, SQL, Git.
• Hands‑on project experience building RAG pipelines, NLP text classification, sentiment analysis, computer vision (YOLO‑based detection & keypoint detection), multimodal deep‑learning frameworks.
• Solid experience in data engineering: multi‑modal dataset building, data cleaning, ETL, dataset QC, database handling (MySQL, PostgreSQL, MongoDB).
• Familiar with statistical methods, exploratory data analysis; hands‑on experience evaluating CV / NLP models with metrics including MAP, NDCG, recall, precision, AUC.
• Working knowledge of visualisation tools: Power BI, Pandas, NumPy, Matplotlib, Seaborn.