|
Mechanical Engineering student at Delhi Technological University I build practical systems with Python and SQL—turning operational data and technical documents into clear, evidence-linked decisions.
|
My work sits at the intersection of AI engineering, data analytics, and engineering workflows. I value interpretable baselines, traceable evidence, reproducible tests, and clear limitations—not just a model that runs.
Start with the decision. Show the evidence. Keep a human in the loop.
My engineering internships provide domain context in turbine maintenance, cryogenic systems, and plant design. The AI projects below are independent portfolio prototypes; they were not commissioned or deployed by those organizations.
|
Decision support for turbine telemetry using data-quality checks, interpretable rolling anomaly detection, transparent maintenance-evidence retrieval, structured recommendations, FastAPI, tests, and evaluation. |
pid-engineering-document-intelligence Typed extraction, engineering-graph validation, revision comparison, and evidence-linked answers for structured P&ID demos, with uncertainty routed to human review. |
|
hr_employee_attrition_analysis Exploratory analysis and classification modeling, translated into business recommendations for understanding employee attrition. |
Operations analysis of 150K ride bookings, including hourly demand patterns, cancellation diagnostics, and recommendations using Python, pandas, and matplotlib. |
Scope note: The AI demos use synthetic telemetry and authored synthetic drawings/procedure excerpts. They are decision-support prototypes—not plant deployments, autonomous control, design approval, or safety certification.
- Start with the user or operational decision, not the model.
- Make assumptions, data gaps, and synthetic-demo boundaries visible.
- Prefer interpretable baselines before unnecessary complexity.
- Test edge cases and keep regression failures reproducible.
- Return evidence, uncertainty, and human-review paths when confidence is limited.
Built around the original Nothing0g character — deep red, black, warm brown, and cream.