Who needs this to work?
Business rules, user workflows, and operational constraints come first. Reconciliation exceptions, workforce planning, and document review each ask something different of a system.
AI ENGINEER · GENAI ARCHITECT · TECHNICAL LEAD
Hi, I’m Naveen. I work hands-on with GenAI, data, and enterprise systems—and with the teams turning them into something useful.
14+ years of engineering, data, and AI.
Experience across the United States and Australia.
CLOUD EXPERIENCE
- AzureOpenAI · AI Search
- AWSSageMaker
- GCPVertex AI
ENGINEERING WITH PEOPLE IN MINDFollow my experience ↓
EXPERIENCE, IN PRACTICE
From data foundations
to hands-on GenAI.
I’ve worked on the systems behind bookings, payroll, workforce planning, and banking. Choose a chapter to explore the responsibilities and technical work in that role.
Artificial Intelligence Engineer with NLP
Building GenAI into
real banking workflows.
My current client engagement brings together AI architecture, hands-on engineering, and technical leadership: connecting enterprise knowledge to LLM applications, automating reconciliation, and working with complex financial documents.
MY RESPONSIBILITIES
- Lead architecture and delivery of enterprise AI, GenAI, NLP, and LLM-powered solutions.
- Define reusable frameworks, evaluate models and tools, and guide cross-functional teams through architecture reviews.
- Design secure enterprise integration, deployment architecture, CI/CD, and monitoring strategies.
A CLOSER LOOK AT THE WORK
01GenAI, RAG & Copilot workflows
+
Designed RAG, semantic search, conversational AI, and document intelligence solutions. Built Copilot Studio topics, actions, API integrations, Power Automate workflows, and custom connectors.
Security work included Entra ID, RBAC, managed identities, Key Vault, prompt-injection defenses, tenant isolation, and tool authorization.
02Automated bank reconciliation
+
Built ingestion, normalization, account mapping, automated matching, exception handling, prior-period item management, and reconciliation reporting.
Python / FastAPI and PostgreSQL services supported approval workflows, period locking, audit history, transaction traceability, and human review.
03Computer vision for financial documents
+
Used OpenCV for quality checks, denoising, normalization, orientation correction, and region extraction. CNN / ResNet models supported classification, quality analysis, and visual-similarity workflows.
Python, PyTorch / TensorFlow, Docker, and AWS / SageMaker supported training and inference. Evaluation covered precision, recall, F1-score, latency, and error analysis.
WORKING WITH
Python / FastAPI · RAG · Copilot Studio · LangGraph / LangChain · PostgreSQL · Azure · AWS / SageMaker
THE CONNECTION ACROSS MY WORK
Good architecture
starts with listening.
My work connects technical decisions with the people and workflows they need to support. These are the questions I bring to an AI project.
What does the AI need to know?
Data quality, retrieval, embeddings, and context management shape the response. I work across the data foundation and the LLM application.
What should it be allowed to do?
Identity, permissions, tenant boundaries, tool authorization, and human review belong in the architecture. So do evaluation and production monitoring.
How do we build it together?
Reusable patterns, architecture reviews, and hands-on technical guidance help AI, data, software, and cloud teams connect their work.
A WEBSITE CAN ONLY TAKE US SO FAR
The next part
is a conversation.
A role, an AI challenge, or an idea for working together. Reach out directly—I’d like to hear what you have in mind.
get2navee@gmail.com ↗Or connect on LinkedIn ↗