Ojas M Agarwal

Optimized data integration and automation processes for GST/TDS, significantly reducing processing and deployment times while enhancing data quality.

• Spearheaded the adaptation of SAP ABAP logic into modular DBT models, achieving a 95% reduction in GST/TDS integration time.
• Developed Python backend scripts to validate GST filing data, decreasing manual QA time by 40%.
• Automated GST filing by transforming Oracle(EBS) ERP data using SQL and DBT, leading to a 70% reduction in manual effort.
• Streamlined CI/CD pipelines by integrating validation modules via GitHub Actions, cutting deployment time by 50%.

Contributed to the development and validation of an NLP pipeline for educational assessments, leveraging advanced AI models to enhance content generation and evaluation.

• Optimized RoboTutor's NLP pipeline to enhance scripted assessments, improving educational content for marginalized children.
• Validated the Neural Symbolic Parser (NSP) as a proxy for human-like reasoning, achieving 88% accuracy with ChatGPT-4o for English.
• Automated NSP JSON generation for 100+ stories using Python and Translate API, effectively handling complex edge cases.
• Evaluated ChatGPT-4o's comprehension by testing over 12,000 NSP questions across two languages using the OpenAI API.

Drove AI-driven analytics and content moderation initiatives, automating reporting and enhancing user engagement through innovative solutions.

• Automated report generation using Pandas and ReportLab, reducing Mixpanel analysis time by 80%.
• Designed PowerBI dashboards to track over 4,000 clicks, revealing key subject-wise engagement and retention patterns.
• Implemented real-time content moderation by deploying AWS Comprehend and Rekognition pipelines, ensuring safety and compliance.
• Integrated MuseTalk for real-time lip-sync capabilities into chatbots, significantly enhancing engagement via video responses.
• Researched and applied PandasAI and SteveAI to advance AI-driven analytics and video automation workflows.

Developed and deployed advanced AI and Computer Vision solutions, optimizing 3D modeling, document processing, and voice command systems for logistics and inspection.

• Constructed a NeRF-based 3D pipeline utilizing NVIDIA Instant-NGP for virtual and remote warehouse inspections.
• Developed and deployed YOLOv5-based APIs on AWS EC2, automating document extraction and reducing manual effort.
• Fine-tuned the Gemini API for precise JSON extraction from packing list images, significantly boosting data accuracy in logistics.
• Integrated Wit.ai with custom intents for voice commands, improving recognition and user interaction by 91%.

Led fundraising efforts for soldier rehabilitation, demonstrating strong organizational and community engagement skills.

• Raised over 90 Lakhs INR (approx. $100,000+) for the rehabilitation of soldiers affected by the Pulwama terror attack.

Contributed to community development by improving sanitation infrastructure for underprivileged children.

• Helped build essential sanitation facilities for underprivileged children in a local government school.

Participated in national events as an NCC Cadet, demonstrating discipline and leadership.

• Selected to march on the Rajpath during India's Republic Day celebrations, showcasing national pride and discipline.

Awarded for presenting published research at the Thirty-Ninth AAAI Conference on Artificial Intelligence (AAAI'25).

Recognized for outstanding academic excellence and consistent high performance during academic years 2016-2019.

Student abstract presented at the AAAI Conference on Artificial Intelligence, focusing on evaluating text comprehension using next sentence prediction in large language models.