AI & Learning Lab

Everything I am building and learning.

Live automations, dashboards, and tools — built from scratch while learning AI and programming. Each project here started with a real problem and ended with something that actually runs.

3
Live Automations
20+
Hours of Admin Saved
100%
Built from Scratch
Personal

Personal Projects

Things I built to solve problems in my own life. Each one taught me something that carried forward into the next project.

Make.com · Gmail · Scheduler

Daily Morning Wellness Reminder

01
Live & Active
Make.com Gmail HTML Email Scheduler
Trigger
Daily at 8:00 AM IST
Output
Personalised HTML email
Effort after setup
Zero
What I Built

An automated daily reminder that sends a personalised HTML email to my inbox every morning at 8:00 AM — with three prompts: soak millet for Ambali, a motivational quote, and a reminder to eat roasted seeds. Built entirely in Make.com with no coding.

Why I Built It

When mornings get busy, small habits slip. Instead of relying on memory or phone alarms, I built a system that delivers a structured nudge automatically — every day, without me touching it.

How It Works
1
Schedule trigger fires at 8:00 AM IST
Make.com's scheduler activates the scenario daily at the exact time, without needing any device to be on.
2
HTML email is assembled with the three reminders
Make.com builds a formatted HTML email with structured headers and content — the same format every morning.
3
Email delivered to Gmail inbox
It arrives before I start my day. I open it, read it, act on it. That is the entire workflow.
What I Learned

How to connect a schedule trigger to a Gmail send action in Make.com, map dynamic content into HTML email format, and build a scenario that runs reliably without any manual input. This was my first ever automation — the one that made everything else feel possible.

Skills This Built

Trigger-based automation logic, HTML email formatting, Make.com scenario structure, data field mapping, and debugging failed runs using execution history.

🏆
Key Achievement
My first live automation. A scheduled, personalised email that runs every morning with zero manual input. Proved to myself that I can build real working systems — not just read about them.
Professional

Professional Projects

Tools built to solve real problems in my day-to-day work — tracking requests, processing meeting notes, and cutting down on manual admin.

Make.com · Gmail · Google Sheets

Gmail to Google Sheets Request Tracker

02
Live & Active
Make.com Gmail Google Sheets Data Mapping Filter Logic
Trigger
Every 15 mins — Gmail watch
Filter
"Request" in subject line
Output
Auto-logged row in Sheets
Columns logged
Date, Sender, Email, Subject, Status
What I Built

An automated request tracking system that watches Gmail every 15 minutes and logs every email with "Request" in the subject line into a Google Sheet — capturing date, sender name, sender email, subject, and a default status of "New".

Problem It Solves

Manually copy-pasting email details into a tracker is time-consuming and error-prone — especially across multiple active projects. This automation captures every request the moment it arrives, creating a clean audit trail with zero human effort.

How It Works
1
Gmail watch triggers every 15 minutes
Make.com polls the inbox on a 15-minute cycle, checking for new emails since the last run.
2
Filter checks subject for "Request" (case-insensitive)
Only matching emails proceed to the next step. Everything else is ignored.
3
Email fields are mapped to Sheets columns
Date, sender name, sender email, and subject are extracted and routed to the correct column.
4
New row added with Status set to "New"
The row is inserted into the tracker instantly. I update the status as I action each request.
PM Relevance

This mirrors real-world workflows — stakeholder request tracking, escalation logging, and ticket intake. The watch-filter-log pattern is the same one that powers JIRA automation rules and helpdesk ticketing systems.

What I Learned

How to connect Gmail to Google Sheets using Make.com, build a filter condition that keeps data clean, and design column structure before building the automation — a habit that directly mirrors how I approach tracker setup in professional projects.

🏆
Key Achievement
A real-time email tracking system that auto-logs all incoming requests to Google Sheets — the same workflow used in professional PM and operations teams. Zero copy-paste. Full audit trail. Runs every 15 minutes without any input from me.
Google Apps Script · Claude API · Google Drive · Gmail

AI Meeting Transcript Processor

03
Live & Active
Google Apps Script Claude API Google Drive Google Sheets Gmail
Trigger
Every 5 mins — Drive folder watch
Input
.docx transcript in Drive
Output
Action items in Sheets + Gmail draft
AI Model
Claude API (Anthropic)
What I Built

An end-to-end AI pipeline that watches a Google Drive folder and automatically processes any meeting transcript uploaded to it. Drop a .docx file in — Google Apps Script reads it, sends the content to the Claude API, and instructs the AI to extract all action items, owners, and deadlines. The results are logged as individual rows in Google Sheets and a formatted follow-up email is created as a Gmail draft, ready to review and send.

Problem It Solves

After every meeting, reading through transcripts to pull out tasks, assign owners, log deadlines, and draft a follow-up can take 20 to 40 minutes. Across multiple meetings a week, across multiple projects, that time adds up fast. This automation reduces the entire process to one step: upload the file.

PM Relevance

Post-meeting follow-up — extracting actions, assigning owners, logging tasks, communicating next steps — is one of the most time-consuming parts of project coordination. This automates it completely, and shows how AI can be embedded into structured PM workflows, not just used as a chat tool.

How It Works
1
Upload .docx transcript to the "Meeting Transcripts" Drive folder
The only manual step. Drop the file in and everything else runs automatically.
2
Apps Script detects the new file every 5 minutes
A time-based trigger scans the folder for unprocessed .docx files using a strict MIME type filter.
3
File is converted to Google Doc and text is extracted
A temporary Google Doc copy is created, text is read, and the copy is immediately deleted to keep the folder clean.
4
Text is sent to Claude API with a structured extraction prompt
Claude reads the transcript and returns all action items, owners, and deadlines as structured JSON.
5
Action items logged to Google Sheets — one row per task
Task, owner, deadline, and status ("Open") are written as separate rows in the tracker.
6
Gmail draft generated with formatted meeting summary
A ready-to-send follow-up email is created as a draft. I review, adjust if needed, and click Send.
7
File renamed to "processed_" to prevent duplicate runs
The script skips any file with this prefix on future runs — each transcript is handled exactly once.
What I Learned

How to build a multi-step automation pipeline entirely in Google's cloud using Apps Script — no local machine, no paid third-party tools. How to call an external AI API, parse structured JSON responses, and route output to multiple destinations simultaneously. I also debugged a significant real-world issue: a missing MIME type filter caused the trigger to process all file types, generating thousands of duplicate files. Diagnosing and fixing this taught me how critical defensive coding and trigger hygiene are — skills that transfer directly to managing automated workflows in any environment.

🏆
Key Achievement
First project to integrate a live AI API call into a real working workflow. Turns a raw meeting transcript into a structured action item log and a ready-to-send email — reducing post-meeting admin time to zero.
— What's Next

More things in progress.

Not all of these will be PM tools. Some will be experiments. Some will solve personal problems. All of them will be real.

Work Tool · In Progress

Automatic Task Chaser

Scans the action item tracker daily and sends reminder emails to task owners who are approaching or past their deadline — without me having to chase anyone manually.

Work Tool · Planned

Auto Stakeholder Report

Every Friday, pulls open tasks from the tracker, summarises project status using AI, and emails a formatted report to stakeholders automatically.

Personal · Planned

Dashboard

A personal tracking dashboard — what it tracks is still being decided. Could be habits, projects, reading, or something else entirely. Building it to learn frontend and database skills.

Learning · Planned

First App

A small standalone app — purpose not finalised yet. The goal is to go beyond automations and build something with a real interface. Learning as I go.

Want to work with someone building this?

I'm actively looking for my next role where I can bring both international delivery experience and an automation-forward mindset to a high-impact team.