Live SEC EDGAR API integration with automated Telegram alerts, validated against real filings.
Black-Edge ↗
SEC Filing Intelligence & Alert System
Live SEC EDGAR integration · automated Telegram alerts
Welcome.
I find where AI can save a business time, help it make more money, or stop important work from falling through the cracks — then I build an operating system that makes it happen.
Start with the problem and the goal, not the technology. Use AI only where it can create a clear benefit.
Harsha E
Medical Doctor — MBBS · MPH, University of Southern California · Greater Toronto Area
Selected Work
Four systems I designed, built, and ran end to end. Each one is running against real data, not a prototype.
I built these for my own use and I am taking on first client projects now, so the work below is the evidence rather than a client list.
End-to-end pipeline: automated posting ingestion, targeted résumé generation, automated quality checks, structured JSON application tracking. Diagnosed a silent-failure bug and scaled output from 18 lifetime results to 828 per run.
Job OS ↗
Automated Job Opportunity Pipeline
828 postings per run · up from 18 lifetime
Installable desktop app with live meeting monitoring, transcript backfilling, saved state, troubleshooting tools, and an executable installer.
Zoom Transcript Capture ↗
Desktop Application
Live capture · backfill · packaged installer
Finds low-cost acquisition and flip opportunities from public feeds. Built on a reusable pipeline architecture: scout → filter → dedup → score → match → compose.
Hermes Digital Workers ↗
Acquisition Scout
Finds low-cost acquisition and flip opportunities from public feeds
Services
Every engagement ends with something running in your business, not a slide deck about what could be built later.
Productized and outcome focused. Both routes start with the audit.
01
AI Opportunity Review
Where AI creates real upside in your business, and where it is a distraction. You get the highest leverage opportunities ranked by what each is worth against what it takes to build.
02
Systems That Pay for Themselves
The build that follows. Every lead answered while it is still warm, a weekly report that writes itself, decisions made on numbers that are current rather than last month. If it will not pay for itself, I will say so.
How it works
Three steps, in this order. Most AI projects fail because someone skipped straight to the third.
01
Where you are
We map the actual process, not a description of it. Where the hours go and where things fall through is usually not where people expect.
02
Where you want to be
We name the outcome in concrete terms. Hours back, leads answered, reports that write themselves, so success is something you can check rather than something you feel.
03
Close the gap with AI
I build the smallest system that gets you from one to the other, and I make failure loud so a break is visible instead of silent. If AI will not close the gap, I say so before you spend money.
Two ways in: client projects, or value based partnerships with businesses where the win is shared.
What the research says
The gains are real and repeatedly measured. Most deployments still miss them.
I am early in client work, so here is the evidence that matters more than my opinion: what measured studies report about applying AI to ordinary business tasks.
14%
40%
26%
4 in 5
That last number is the reason I lead with a review rather than a build. The gains are real and repeatedly measured, but they arrive only when the work is pointed at a process that actually costs something. Choosing correctly matters more than the model you use.
About Me
Fifteen years in medicine before I ever wrote a line of production code.
Harsha E
My name is Harsha. I trained as a physician in India, took a Master of Public Health at USC, and worked in clinical medicine, research, and public health before moving into investing and operations.
For the last few years I have been building AI systems full time. Systems that run on a schedule, handle real data, and get checked when they break. Job pipelines, regulatory filing screeners, meeting capture tools, and lead generation agents, each built because I needed the thing to exist and nobody had built it.
The useful part of the medical training is not the clinical knowledge. It is the habit of asking what happens when this fails and nobody notices. That is where most AI projects quietly die, and it is exactly where I focus.
It also means I read healthcare and clinic operations from the inside. Intake, scheduling, referrals, billing follow up, and the paperwork around all of it are workflows I have lived in, which most people building AI tools have only read about.
2024
AI Systems Builder
Independent
2019
Investing and Operations
Real Estate and Private Holdings
2016
Medical Liaison
Sonologix
2010
Medical Officer
Texas Chemtex
2009
Clinical Research Associate
City of Hope
2007
Master of Public Health
University of Southern California
1999
Medical Doctor — MBBS
Rajiv Gandhi University of Health Sciences
Tell me where the bottleneck is — what’s slowing the work, creating leakage, or limiting revenue. I’ll tell you whether AI is a practical way to remove it.
Get in touch →No pitch deck, no obligation. A short conversation about the actual problem.