I’m an AI engineer. I connect models to the data, tools, and interfaces that make them useful in the real world.
How I got here
At Ujyalo Ventures, I worked on turning business questions into checked queries and visual answers.
As Lead AI Engineer at BusinessFrame, I worked across a larger product: analysis, agents, campaign workflows, and the decisions people need to control.
Now I work independently, building AI products and automations from the interface through to the infrastructure.
Here’s what that looks like.
Five systems. Different problems. The same attention to the whole product.
BusinessFrame2025 — 26
A website becomes a campaign plan.
As Lead AI Engineer at BusinessFrame, I worked across website analysis, strategy reports, agent workflows, and Google Ads Search campaign provisioning.
The recording follows website analysis, service signals, and keyword filtering.
Campaigns can be prepared in a paused state while billing, conversion tracking, and policy checks go through human review.
Next.js / AI agents / Google Ads API
OpsPilot2026
An inbox becomes a workflow.
I built OpsPilot to turn operational email and documents into auditable cases: intent, extracted facts, policy evidence, and a proposed action in one place.
Watch incoming email become a classified case with policy evidence and a proposed action.
Commissioned work, demonstrated with synthetic data.
Agent workflows / RAG / Human review
Text-to-BI2024 — 25
A question becomes an answer you can inspect.
At Ujyalo Ventures, I worked on an agentic BI assistant that selects schema context, generates SQL, and streams tables or visualizations in response to business questions.
Follow the document from upload to structured fields and human review.
The original image stays alongside the record so the extracted data can be checked.
Vision OCR / Next.js / SQL
And sometimes, it starts with a challenge.
AI IELTS Prep put speech recognition, an LLM examiner, and generated speech into a speaking-practice experience. Built in 48 hours. Winner of AI Crusade 2024.