AI/ML Developer at WARRGYIZMORSCH (2024-05 – Present)
Developed Python-based ML pipelines for educational content analysis, automating quality scoring and content tagging using Scikit-learn classification models, reducing manual review time by 35%.
- Developed Python-based ML pipelines for educational content analysis, automating quality scoring and content tagging using Scikit-learn classification models, reducing manual review time by 35%.
- Built a REST API wrapper around a text classification model to serve content-type predictions to the internal platform, integrating it with the existing curriculum management workflow.
- Implemented NLP-based keyword extraction and readability analysis on 500+ lesson documents using NLTK, enabling data-driven content structuring decisions.
- Designed a data validation pipeline using Pandas with custom rule engines to flag inconsistencies in lesson metadata, achieving 95%+ anomaly detection prior to content publication.
- Containerized and deployed ML services on AWS using Docker and EC2, integrating CI/CD pipelines for seamless updates and scalable production deployment.
- Experimented with prompt engineering on LLM APIs to auto-generate lesson summaries and learning objective suggestions, reducing content drafting time by 20%.
- Collaborated with 10+ clients to understand requirements, translated them into ML problem statements, and delivered working solutions with documented results within agreed timelines.
- Participated in code reviews across 10+ deliverables, improving solution quality and consistency across the team.
Data Analyst at NPCIL (Nuclear Power Corporation of India) (2023-05 – 2024-05)
Developed automation scripts in Python to streamline data entry and processing, reducing manual workload by 40%.
- Developed automation scripts in Python to streamline data entry and processing, reducing manual workload by 40%.
- Designed and implemented a data validation system using Pandas to ensure the accuracy of computerized records.
- Created Python-based ETL (Extract, Transform, Load) pipelines to automate data movement between internal databases.
- Built custom reporting dashboards using Matplotlib and Seaborn, enabling better visualization of key operational metrics.
- Integrated OCR-based document processing with OpenCV to digitize and manage inward and outward dispatch records.
Dradis Moderator at ARCGATE (2022-05 – 2023-05)
Conducted 25,000+ account moderation reviews using the Basestar tool, mitigating 95%+ flagged accounts for user safety.
- Conducted 25,000+ account moderation reviews using the Basestar tool, mitigating 95%+ flagged accounts for user safety.
- Performed 10,000+ job-level and account-level reviews, addressing 90%+ detected fraudulent activities and ensuring platform integrity.
- Scanned and acted on 12,000+ suspicious accounts and content, maintaining high user safety standards. Collaborated with 5+ cross-functional teams to refine moderation guidelines, reducing response time to fraud patterns by 30%.
- Analysed user data to identify 30+ recurring fraud trends, implementing strategies that reduced fraudulent incidents by 25%.
- Documented findings and contributed to 15+ process optimizations, enhancing account safety protocols in the Dradis system.
