AI BUILDER & STRATEGIST
I'm a visionary AI builder and strategist with an innovative approach to bridging technology, data, and business. I envision AI solutions that transform possibilities into realities, crafting intelligent systems that pioneer new frontiers in human-AI collaboration.
My three superpowers:
- • Analytical rigor
- • Hands-on AI engineering
- • Proactive communication style + people-connecting
What people say
A few words from collaborators and leaders I've worked with.
"Nana is an expert in her field and constantly pushing the boundaries of what can be done. She's quick to help others and offers good advice. She is very smart and has a strong desire to be better at her job. She asks for feedback often and makes changes based on it."
Supervisor
"Nana is the real starter who thinks outside of the box and challenges others. She is remarkably intelligent and a great coworker. She is the one person on our team willing to ask hard questions, and I seen it as a quiet but confident leader among her peers."
Colleague
"She very well understands her professional domain and opportunity for contribution to the wider organization. She communicates problems in a clear and thoughtful way - careful to understand the general so she doesn't narrow context. She is transparent with her work and thinking and regularly seeks input, earnestly incorporating feedback into her work and process."
Colleague
"She cares about understanding the business needs and effectively collaborating with business stakeholders to ensure a data science solution is actionable."
Colleague
"Nana's energy and attitude always inspired people to push harder, no matter how big or small the problem or idea."
Colleague
Experiments & systems I've shipped
Prototypes that proved real impact
LLM-powered Call Resolution Analytics and Agent Coaching Tool
Enterprise-level AI system using LLM to analyze raw call transcripts, measure call resolution effectiveness, and distill actionable 'why' insights for customer support teams and agent coaching programs.
RAG-powered CX Intelligence Assistant
Led innovation of a RAG-powered CX Intelligence Assistant to democratize insights across research and CX. Built a CX research chatbot that reduced qualitative analysis against customer surveys from days to minutes, using hybrid search architecture with cross-encoder reranking and a metadata-aware query rewriting layer for higher precision and retrieval relevance.
Linda - Your Daily Health Companion
The first AI assistant that listens to your days, spots personal triggers for chronic pain, and coaches you towards fewer and milder attacks. Built around three core actions: Talk – foregrounds the voice-first, conversation-driven flow; Track – reassures data seekers that you still log and analyse; Tame – promises action and improvement, not just monitoring.
Writing
You've Got a Shiny AI Hammer—But Are You Hitting the Right Nails?
Read article →Lab notes
Thinking out loud, theories in progress, and questions I'm exploring.
Beyond Notebooks: Text-to-SQL Agent Failure Modes & Reflection Loops
Studying how a simple text-to-SQL agent fails reveals that the real issues aren't syntax errors but contract problems — incomplete schema semantics, reasoning shortcuts, and lost traceability. Argues that output contracts and lightweight validators beat prompt rules for building durable, model-agnostic agent systems.
Read on LinkedIn →Beyond Notebooks: Docker Debugging & Abstraction Layers
A Docker debugging session revealed that a PostgreSQL authentication error was actually a port conflict hidden across abstraction layers. Explores how studying AI's reasoning path can sharpen our own engineering thinking.
Read on LinkedIn →Let's build something interesting
I'm always interested in collaborating on ambitious AI systems, experimental prototypes, and projects that push what's possible. Reach out if you have something we should talk about.