💳 Loan Eligibility Checker App
Open ↗
Data Scientist • AI/ML Engineer • GenAI + Azure ML • 2× Inventor
6+ years across insurance, banking, and healthcare. GenAI (LLMs, RAG, LangChain), predictive modeling, and MLOps on Azure & AWS.
Certified Data Scientist, Inventor, and AI Researcher specializing in AI security and governance. With 6+ years of project experience across insurance, fintech, and healthcare, I have designed and deployed end-to-end AI/ML solutions that combine Generative AI, NLP, and predictive modeling with real-time MLOps on Azure and AWS. My work includes filing patents on AI governance frameworks and building production-grade systems that deliver measurable business impact while ensuring compliance and responsible AI practices.
Multi-agent orchestration and oversight framework for LLMs, enabling persistent interpretability, explainable meta-reasoning, and cross-perspective decision audits. Focused on regulatory alignment and safe autonomy in agentic AI systems. (Provisional)
Real-time compliance monitoring system using GPT-4, FAISS, chunked embeddings, and RAG. Automatically identifies potential policy violations and generates context-aware alerts through adaptive feedback mechanisms. (Provisional)
SentraNova is my research initiative for building controlled agentic AI systems that remain interruptible, auditable, and policy-aligned under real-world constraints. It introduces a governance-first orchestration layer for LLM agents—balancing autonomy with oversight to ensure reliable, compliant decision-making.
SentraNova orchestrates task agents, critic agents, and a governance layer over tools/APIs. Retrieval (RAG), vector checks, and policy evaluators run in-loop. Every step emits structured telemetry for observability (metrics, traces, decisions) and supports real-time interrupts, human-in-the-loop overrides, and safe fallbacks.
Governance Layer
Policies, role-based constraints, action gates, approvals.
Agents & Critics
Task agents propose; critic agents verify risk, bias, and compliance.
Telemetry & Audits
Signed logs, replay, counterfactuals, and drift monitors.
I'm actively evolving SentraNova toward reference implementations across claims processing, risk decisions, and clinical support—bringing safe autonomy to real-world AI systems.
Focused on Generative AI, scalable MLOps, and cloud-native deployments.
LLMs, RAG, LangChain, Prompt Engineering, OpenAI API, Hugging Face, Vector databases
XGBoost, Random Forest, Time-Series (ARIMA, Prophet), Decision Trees, PCA, NLP
TensorFlow, Keras, CNN/RNN/LSTM, LayoutLMv3, TrOCR
Azure ML, AWS SageMaker, FastAPI, Docker, CI/CD (GitHub Actions, Azure DevOps)
Python (Pandas/NumPy), SQL, Data Factory, Data Lake, Purview
Power BI, Tableau, Plotly, Dash, Streamlit
Current credentials aligned to Azure-first, multi-cloud MLOps.
Azure ML, experimentation, model training, deployment, and monitoring.
Azure OpenAI, Cognitive Services, orchestration, and responsible AI.
Core Azure data concepts, relational & nonrelational stores, analytics workloads, and services like Azure SQL, Cosmos DB, Synapse, and Power BI.
ML and deep learning foundations, TensorFlow/Keras modeling, MLOps basics, and deploying AI solutions with scikit-learn pipelines.
Python and SQL for data analysis, data wrangling and visualization, introductory machine learning, and an industry-style capstone project.
LLM fundamentals and architectures, prompt engineering, RAG with vector databases, fine-tuning/PEFT, safety and evaluation guardrails, and deploying GenAI apps on cloud using frameworks like LangChain and OpenAI APIs.