Skip to content
An ocean wave drawn up and held at the moment just before it breaks

About

I figure things out, move fast, and close what I start.

I build agentic AI systems that run in production for real companies. Five of them so far, across deal flow, lead qualification, outreach and content, and they are all still running.

Resume
Illustrated portrait of Pari Goyal with a root system spreading outward behind her

I think in conversations, move in decisions, and execute in code. I'd rather ask the dumb question in the room than ship the wrong thing.

Experience

Four rooms, four problems, the same instinct.

First hire · Akroventures

Apr 2026 — Present

Built the firm's entire digital infrastructure from nothing in three months.

  • Built a deal-management CRM on Next.js, Supabase and the Claude API: public intake, CSV bulk upload, AI-assisted deal analysis, hot/warm/cold classification, and a tokenized human-in-the-loop approval workflow. It has processed 200+ startup leads and carried outreach behind a fundraising mandate worth up to ₹20 crore.
  • Caught corporate email scanners auto-triggering approval links by pre-fetching them during phishing checks, and fixed it by moving approval from a GET request to an authenticated POST confirmation.
  • Built an n8n content pipeline that turned a daily scramble into a ten-minute review and saved close to 10 hours a week, then fixed the site's search visibility with per-route metadata, JSON-LD and an llms.txt.

Agentic AI Intern · Oyelabs

Mar 2026 — May 2026

Built agents other people ended up depending on.

  • Built AI agents for live client projects on n8n, Supabase, and large language model APIs.
  • Designed a global prompt framework that cross-checked SOPs before any workflow ran, saving the manager a full day every week.
  • Owned prompt engineering across several client projects and kept handoffs clean enough to work the first time.

Marketing & AI Automation · Halo Interiors

Nov 2025

Took the manual work out of video, then wrote it all down.

  • Automated cross-platform video processing with AI workflows and cut out most of the manual steps.
  • Stress-tested Convex, Supabase, and Firebase to find what would actually hold at scale.
  • Documented what broke and what worked, so the next person did not start from zero.

Product Intern · ScoutNexa

Feb 2026 — Mar 2026

Shipped fast, talked to users, learned how early stage really moves.

  • Researched competitors across Crunchbase to sharpen the product problem statement.
  • Shipped an MVP in a few weeks with no-code tools, talked to early users, and iterated quickly.
  • The founder shut the company down, which taught me more about focus and speed than a clean win would have.

A seat where thinking and execution both count.

The role

Agentic AI, AI automation, or founding engineer

The company

Small, AI-forward, moving fast. Remote, anywhere

The bar

Where I build the AI layer and own what it does

Targeting ₹14–20 LPA. If the work is real, I am flexible on the rest.

Broken ceramic pieces rejoining with the cracks filled in gold, in the style of kintsugi

Like kintsugi, the things I have broken and rebuilt are the parts I am proudest of.

A small paper boat at the tideline, partly dissolving into the sand with ripples trailing behind it

What early stage taught me

The boat still left ripples.

Speed is a feature

A rough MVP in front of real users beats a polished idea nobody has touched. The feedback is the product.

Documentation is a gift

Writing down what broke and what worked is a gift to the next person, including the version of me six weeks from now.

A shutdown is not a stain

When the founder closed ScoutNexa, I learned more about focus and pace than a clean win would have taught me.