Hi, I'm Ishika.
enterprise trained, startup tested
Listen to today's AI news in under 10 minutes, curated by Ishika's agentic AI system, while you learn more about me.
An agentic AI system that scouts, fact-checks and narrates a daily AI news podcast on its own. Built solo on the Claude Agent SDK. Runs free, on a schedule, no server.
Not one model call in a costume. A newsroom of six specialized agents that hand work to each other and argue before they publish. A Fact-checker can send the Writer's draft back for a second pass. That loop is what makes the output trustworthy enough to actually send.
The honest part: the agents were the easy twenty percent. Every real fight was in the glue. The free voice that worked on my Mac and got blocked from the cloud. The run that reported success and silently sent nothing. That is the real story of shipping AI.
Product teardowns, build logs and honest reflections.
WalletHub has built a genuinely differentiated product. The homepage undersells it. The biggest opportunities aren't visual. They are making it obvious who the product is for and helping the brand feel more human.
I vibe-coded StockPulse, an S&P 500 dashboard, to test the new AI building tools. The honest report: the tools made the code cheap and fast, and left every decision that actually mattered to me.
An honest account of preparing for, going through, and learning from a top-tier consulting interview process. What the case room teaches you that no framework does.
An honest look at what an MBA actually buys you in 2026, when AI can run the analysis and the network is a LinkedIn search away. What the degree gave me that the internet couldn't.
Two Claude skills I built and use myself. Install them in one command.
Paste a job description and your resume. The skill rewrites your bullet points to match the role, in your voice, without inflating anything.
Get it on GitHub ↗Draft a cold email to a hiring manager or recruiter. Input the job, the person, your background. Output: a short, direct message that doesn't sound like ChatGPT wrote it.
Get it on GitHub ↗Standout individual pieces on product, supply chain tech and leadership.
A practical breakdown of agent loops a PM can run this week, and how to define what "done" means so they stop on their own.
Fifty ways non-technical people put Claude Code to work. It is why I stopped thinking of it as a coding tool.
How last-mile carriers and delivery platforms are reshaping parcel logistics. The operational side of the problems I want to build for.
Why lasting change is neurological rather than a matter of willpower, and what that means for building hard habits.
Public speaking as a learnable career skill, with simple frameworks for impromptu answers and the exec room.
enterprise trained, startup tested
I am a product manager who builds. Five years shipping enterprise software, an MBA from Michigan State's Broad College of Business, and a habit of writing the code myself when an idea is worth proving. Plenty of PMs can describe an agentic AI system. I have built one that runs every morning without me.
I learned product at Siemens, inside a two billion dollar software portfolio. I took a failed feature from zero traction to 31% adoption by going back to customers and redesigning the roadmap around what they actually did, and I ran an Agile transformation that removed 15,000 engineering hours of rework a year. Big organizations taught me how decisions really get made, and how much patience it takes to move one.
The last two years I have spent at the front edge of AI, building. AI Newsroom is six agents on the Claude Agent SDK that scout the day's news, fact-check their own drafts and narrate a podcast, running free on a schedule with no server. I vibe-coded a live S&P 500 dashboard to find out what the new tools hand you and what they quietly leave on your desk. The answer, both times, was judgment.
The through-line has not changed since I wrote my first line of code. Find the real problem. Build the thing that solves it. Know when to stop. I am looking for a team building something hard, where a PM who can talk to engineers and executives in the same afternoon is exactly what the work needs.