Anushka Singh
Bangalore
Overview
1
1
year of professional experience
Work History
Data Scientist-I
SIXT Research and Development
Bangalore
06.2025 - Current
- Deploying the chargeback prediction model into production to improve operational efficiency in car rental processes.
- Developing an AI-driven damage detection and employee performance analytics system, identifying anomalies in performance.
- Conducting root-cause analysis on anomalies and providing actionable insights to branch managers.
Data Science Intern
SIXT Research and Development
Bangalore
03.2025 - 05.2025
- Built chargeback prediction system using structured data and ML models.
- Contributed to feature engineering, EDA, and pipeline development.
Education
Master of Science by Research - Computer Science and Engineering
Indian Institute of Technology, Madras
01.2025
Bachelor of Technology - Computer Science and Engineering
Madhav Institute of Technology and Science, Gwalior
01.2022
Skills
- Languages: C/C, Python
- Libraries/Tools: PyTorch, HuggingFace, Scikit-learn, NumPy, Pandas, LaTeX, GitHub, Wandbai, Django, Vscode
- Coursework: DSA, OS, OOP, DBMS, Linear Algebra, ML, DL, CV
- Interests: NLP, Generative AI, Deep Learning
- Analytical problem solving
- Machine learning
- SQL databases
- Statistical analysis
- Scikit-learn
- Natural language processing
- Neural networks
- Feature engineering
- Anomaly detection
- Linear algebra
- Deep learning
Research Activities (MS Thesis)
- Machine Translation Evaluation for Low-Resource Indian Languages Jul 2023 – May 2025
- Analyzed linguistic diversity and data gaps for Assamese, Kannada, Maithili, and Punjabi, and developed a benchmark dataset with 1,000 MQM annotations for zero-shot MT evaluation.
- Enhanced COMET models via Indic fine-tuning and replacing XLM-RoBERTa with IndicBERTv2, improving correlation with human judgments.
- Generated 100k+ synthetic sentences to boost multi-stage training strategies and improve model performance on low-resource languages.
- Adapted GEMBA-MQM (GPT-4) and leveraged LLMs like GEMMA and LLaMA for reference-free, zero-shot evaluation, creating diverse synthetic data to enhance error detection and correction in MT outputs.
Positions of Responsibility
- PG Placement Coordinator, Training and Placement Cell, IIT Madras (2022–25)
- Sports Secretary, Tunga Hostel, IIT Madras (2023–24)
Publications
- How Good is Zero-Shot MT Evaluation for Low-Resource Indian Languages? (ACL 2024) Anushka Singh, Ananya B. Sai, Raj Dabre, Ratish Puduppully, Anoop K., Mitesh Khapra
- Quality Estimation and Post-Editing Using LLMs For Indic Languages: How Good Is It? (MT Summit 2025) Anushka Singh, Aarya Pakhale, Mitesh Khapra, Raj Dabre
Timeline
Data Scientist-I
SIXT Research and Development
06.2025 - Current
Data Science Intern
SIXT Research and Development
03.2025 - 05.2025
Bachelor of Technology - Computer Science and Engineering
Madhav Institute of Technology and Science, Gwalior
Master of Science by Research - Computer Science and Engineering
Indian Institute of Technology, Madras