Statistics Specialist

About the Role At Innodata, we’re working with the world's largest technology companies on the next generation of generative AI and large language models (LLMs). Key Responsibilities Design and implement statistical frameworks for the preparation, analysis, and quality assessment of large-scale datasets used in training LLMs. Perform exploratory data analysis (EDA) to detect biases, data inconsistencies, and anomalies in training corpora. Collaborate closely with AI researchers and engineers to develop data annotation guidelines and evaluation protocols. Apply advanced statistical methods to assess model outputs and guide iterative training improvements. Develop and maintain automated tools for data preprocessing, labeling, sampling, and quality control in LLM pipelines. Conduct rigorous hypothesis testing, A/B testing, and performance benchmarking of LLMs across various tasks. Interpret model behaviors using statistical insights and provide actionable recommendations for model fine-tuning. Document methodologies, maintain reproducible workflows, and contribute to technical publications. Key Qualifications Master’s or PhD in Statistics, Applied Mathematics, Data Science, Computer Science (with strong statistics focus), or related quantitative field. Strong knowledge of statistical modeling, probability theory, and experimental design. Hands‑on experience with large‑scale datasets, data cleaning, feature engineering, and statistical analysis tools. Proficiency in programming languages such as Python, R, or Julia, including libraries like pandas, NumPy, SciPy, scikit‑learn, or statsmodels. Familiarity with machine learning frameworks (TensorFlow, PyTorch) and understanding of LLM architectures (transformers, attention mechanisms). Experience with annotation tools and data labeling processes is a plus. Strong analytical thinking, attention to detail, and problem‑solving skills. Excellent written and verbal communication skills in English. As part of the project, you are required to complete the English language assessment. Assessment The assessment is mandatory & non‑billable. If interested, kindly share your updated resume at: ****** #J-18808-Ljbffr