AI/ML Engineer building production-grade systems — calibrated ML pipelines, LLM reasoning layers, and high-concurrency APIs with explainability, failure handling, and real-world deployment.
- Clinical AI systems with SHAP explainability, RAG reasoning, and HITL governance
- High-concurrency ML inference APIs — 160 RPS · sub-150ms latency · load tested
- Multi-agent LLM pipelines with LangGraph orchestration and failure isolation
- End-to-end ML systems — training → versioning → deployment → drift monitoring
| System | Metric |
|---|---|
| Transaction Risk API | ~160 RPS · sub-150ms latency · 0% failure rate under load |
| CKD Clinical AI Pipeline | AUC 0.999 · 166ms RAG retrieval · 3.6s end-to-end |
| Explanation Alignment | 100% SHAP feature coverage · 0.80 reliability score |
| Robustness Analysis | AUC degradation quantified (0.999 → 0.66) under distribution shift |
| Layer | Tools |
|---|---|
| ML & DL | XGBoost, Scikit-learn, SHAP, PyTorch, Transformers |
| GenAI & LLMs | RAG, LangChain, LangGraph, Prompt Engineering, Guardrails |
| Backend & Infra | Python (Async), FastAPI, Docker, PostgreSQL, CI/CD (GitHub Actions) |
| MLOps | Model Monitoring, Drift Detection (PSI), Model Registry |
| Cloud | AWS , Azure , Render , Google Cloud |