Aramis Lopez
NYSE
Portfolio & résumé New York · 2026

Business analyst at E*TRADE from Morgan Stanley and a 2026 economics graduate of Williams College. On my own time I build software that keeps running without me: a subscription product, data pipelines that publish every day, and research tools that know when not to trade.

Now Business Analyst, E*TRADE from Morgan Stanley · Jersey City, NJ

Fig. 1 Commits, cumulative Apr 14 – Oct 3, 2026
Written by me Made by automations I built Across 19 repositories, GitHub and local branches.
Begin

01

In brief

Business analyst at E*TRADE from Morgan Stanley with a background in economics, commercial analysis and workflow automation. I build Python and AI-assisted tooling to remove manual steps, strengthen data controls and turn operational data into decisions.

Now
Business Analyst, E*TRADE from Morgan Stanley · since Sep 2026
Education
B.A. Economics, Williams College, 2026 · GPA 3.8 · Dean's List
Before
S&P Global · AlphabloQ · HOPE Financial · Euler
Toolkit
Python, SQL, Advanced Excel, Power BI, Salesforce, financial modeling, data visualization
Languages
English and Spanish, both native
Based
New York, NY
Open to
Analyst, operations and business development roles

Selected figures

  • RFP turnaround after my Python automation at S&P Global, down from several hours< 1 hr
  • Business units I presented financial analysis to4
  • Stakeholders on the weekly reporting I consolidated15+
  • Client accounts managed at AlphabloQ30+
  • GPA, Williams College3.8

02

Selected work

Thirteen things I built in 2026, most of them still running. I scope and design each one and write the code with AI coding tools, including Claude Code. The repositories are private, so the figures below come straight from their commit history.

No. 01 · Flagship · firstrung.xyz Live

FirstRung

Job boards say entry-level, then quietly ask for three years. FirstRung reads every full posting and sends new graduates only the roles they can actually land.

Plans
$12/mo · $39/mo · $29 for 90 days
Commits
52 by hand · 34 by its publisher
Code
8,695 lines · JavaScript, Python, HTML, SQL
Since
Jul 22, 2026

i.

The problem

Titles say “entry-level.” The fine print says three years. New graduates spend weeks applying to roles that were never theirs to get, and no job board filters on what the description actually requires.

posting-a.txtEntry level

Junior Operations Analyst

  1. About the role
  2. Support the operations team with reporting, reconciliations and process fixes.
  3. What you bring
  4. 3+ years of experience in operations or finance.
  5. Advanced Excel; SQL a plus.
Rejected3-year floor
posting-b.txtEntry level

Operations Coordinator

  1. About the role
  2. Keep shipments, vendors and schedules moving.
  3. What you bring
  4. No prior professional experience required.
  5. Recent graduates are encouraged to apply.
Verified0 years
Fig. 2Same title, opposite reality. An illustration of the check.

ii.

The check

FirstRung reads the whole description, not the title. A role passes only if nothing in it sets an experience floor. Soft floors like “two years preferred” fail too, and every pass shows the lines that earned it.

firstrung.xyz · full-description check
FirstRung's full-description check: highlighted lines stating no experience is required, and a Verified verdict
Fig. 3From the live site's walkthrough: a role passes with the lines that earned it.

iii.

The product

Members get a daily email, a dashboard of every verified role, and an application tracker. Stripe handles three plans. A free sample shows three real roles to anyone who asks, behind bot checks and rate limits.

firstrung.xyz
FirstRung homepage: Entry-level jobs that are actually entry-level
FirstRung application tracker with columns for new, applied, interviewing and offer
Fig. 4The homepage, and the tracker every member gets.

iv.

The machine

It runs itself. A daily pipeline harvests, verifies, deduplicates and delivers, and five scheduled agents cover the rest: an ops report, onboarding, source health, nightly Stripe reconciliation and a scout that finds new job boards. No language model sits anywhere in that path.

  1. 01HarvestCompany job boards on Greenhouse, Lever, Ashby, Workday
  2. 02VerifyReads the full description for any experience floor
  3. 03DeduplicateA per-member ledger: no role is ever sent twice
  4. 04DeliverDaily email, dashboard, application tracker

Scheduled agents, none of them AI

  • Ops reportdaily
  • Onboardingevery 6 h
  • Source healthdaily
  • Stripe reconcilenightly
  • Board scoutweekly
Fig. 5The daily pipeline. Counts from a September run: 823 candidates in, 403 verified.

Built with

Cloudflare Pages & Functions · D1 (SQLite) · Stripe Checkout with signature-verified webhooks · GitHub Actions · Resend · Python standard library

Language mix

Visit firstrung.xyz

Also built

Open any row for the full note. The strip on the right is each build on one shared calendar: ticks are commits per week, bars are build windows for projects that never lived in git.

Products

RentRadar watches rental listings in New York, San Francisco, Los Angeles, Washington, Miami, San Diego and Chicago, and emails renters as soon as something fits their budget and taste.

I prototyped it with scrapers, then rebuilt it on licensed API data (RentCast, with HUD and Census context) because the big listing sites forbid scraping. The rebuild also let it run entirely on Cloudflare's free tier.

Stack
Next.js 15, React 19, OpenNext on Cloudflare Workers, D1, Stripe, magic-link sign-in, Resend
Hosting cost
$0 a month
Built
June 2026
rentradar.awrnis.workers.dev
RentRadar homepage with a city, size and budget search and a featured Harlem listing

Diurna publishes itself. Tuesday to Saturday it writes a morning brief on the previous session, Sunday a week in review, Monday a long-form theme, plus a chart of the day drawn from 29 FRED series and an economic calendar assembled from BLS, BEA and Federal Reserve schedules.

The rule that matters most: if an official source is down, it publishes nothing rather than stale numbers.

Data
FRED, U.S. Treasury, ECB, exchange closes
Runs
Four times a day on GitHub Actions
Commits
8 by hand · 46 automated
Code
3,783 lines, zero dependencies
Revenue
Display advertising
diurnafinance.xyz
Diurna Finance front page: a week-in-review headline beside a live markets table

A studio brand selling premium websites to New York restaurants, clinics, firms and shops. To show range before the first client, I designed and built six demo sites for different trades, each with its own identity.

Demo builds
Restaurant, med spa, law firm, gym, bakery, SaaS startup
Stack
Hand-written HTML, CSS and JavaScript on Cloudflare Pages
Built
July 2026
mercer-studio.pages.dev
Mercer Studio homepage: “Your business is exceptional. Your website should prove it.”
Demo site: Ember & Oak, a restaurant
Demo site: Véla Aesthetics, a med spa
Demo site: Calderwood Partners, a law firm
Demo site: Ironbound, a gym
Demo site: Marlow & Grand, a bakery
Demo site: Northsignal, a SaaS startup
The studio site and its six demo builds.

Research & analysis

Four estimation questions, word for word from Many Labs 1, the 2014 project that re-ran 13 psychology effects with more than 6,000 people. Each question opens with a number, and a coin flip on the server decides whether it is low or high. After the fourth answer the page reveals the trick and plots everyone's guesses live: low-anchor answers above the line, high-anchor answers below.

The design, exclusions, stopping rule and analysis were registered and archived before the first response. The analysis copies the original: answers outside the two anchors do not count, then a rank-based t-test and Cohen's d for each question, set against the published effects.

Status
Collecting: 0 of 200 counted, closes Nov 17, 2026
Registered
Archived Oct 3, 2026, before the first response
Stack
Cloudflare Pages Functions, D1, vanilla JavaScript and SVG
Commits
2 by hand
Code
1,452 lines
ballpark-game.pages.dev
Ballpark question: Mount Everest is taller than 2,000 feet. How tall do you think Mount Everest is?
A question as participants see it, here with the low number.

Automation & data

The engine collects every posting from company job boards across 11 applicant-tracking systems, keeps business-side roles in the US or remote, then reads each full description and rejects anything that asks for more than two years.

It is the source FirstRung builds on, and it is tested like one: 438 automated tests cover the parsers, the experience gate and the ledgers that keep a role from ever being shown twice.

Commits
43 by hand · 52 automated
Code
21,351 lines of mostly Python
Schedule
Daily, unattended

Latest sweep · Sep 27, 2026 · 210 seconds

  1. 96,775postings read from 3,009 boards
  2. 9,696match a target title 10.0% kept
  3. 3,174in the US or remote 32.7% kept
  4. 1,531pass the experience gate 48.2% kept
Bars are drawn to scale against every posting read. Figures from the sweep log.

Client acquisition for Mercer Studio. It pulls businesses that list an email but no website from OpenStreetMap across ten US metros, probes every domain and skips anyone whose site already works, then sends a capped, human-paced sequence.

At most 30 emails a weekday, 45 to 120 seconds apart, one follow-up after five days, and an inbox check first so nobody who replied gets a second note. Opt-outs are permanent.

Ran
July 2026, now paused
Commits
10 by hand · 17 automated
Code
453 lines of Python in one file

Every lead it has touched

  • 1,368 businesses screened in 10 metros
  • 1,068 skipped because their website already worked
  • 300 contacted

Markets research

A bot for Kalshi's 15-minute Bitcoin up-or-down markets, 96 windows a day. Before risking money I backtested it across 1,319 windows. Every entry type lost money, and the calibration check showed why: the model was overconfident in every bucket.

Live trading stays locked behind an explicit override. Deciding not to ship it was the useful result.

Trades
1,188
Hit rate
39.2%
Expectancy
−8.52% per dollar staked
Max drawdown
$147.99 on $1,398.69 staked
Code
2,987 lines of Python

Calibration

Expectancy with the best trades removed

  • All trades−8.52%
  • Top 1 removed−9.38%
  • Top 3−10.63%
  • Top 5−11.34%
  • Top 10−13.02%
From the 14-day backtest report. A real edge survives losing its best trades; this one only got worse.

PolyWhale finds skilled Polymarket traders and flags the moments when several of them agree. Most of the work is deciding who to exclude. The top all-time wallet turned out to be a market maker with a −3.6% return on its positions, and 472 of 1,207 candidates were cut for the same reason.

Speed disqualifies too: copying a trader whose median hold is 3.6 hours with a 90-second poller only buys their exits. Sizing uses a quarter-Kelly stake, and on its first live cycle it correctly declined both signals it found because the price had already run.

  • At least 25 closed positions
  • Return on capital of 6% or more
  • Volume no more than 40× profit
  • Median hold of 6 hours or more
  • No more than 55% of profit from one event
  • At least 30% outside sports
  • Traded within the last 10 days
  • At least $5,000 at stake
Code
3,116 lines, standard library only

1,207 candidate wallets, one dot each

  • 472 market makers
  • 654 failed another gate
  • 41 passed, outside the top 40
  • 40 tracked

Apps, games & commerce

A native Android hydration tracker: an animated ring toward a daily goal, one-tap drinks, streaks, badges and weekly analytics. Kotlin with Jetpack Compose, Room for storage and DataStore for settings.

The signed release bundle and the Play Store assets are ready; the listing is next.

Stack
Kotlin, Jetpack Compose, Room, DataStore, AdMob
Code
2,136 lines
Built
April 2026
Water Tracker analytics screen
Water Tracker home screen with a hydration progress ring at 57 percent
Water Tracker achievements screen
Screens from the Play Store listing assets.

A calendar of the month with a list for every day. Finished tasks sink to the bottom, yesterday's leftovers come over with one click, repeats follow daily, weekday or weekly rules, and a streak only counts days that had something planned.

It runs on a laptop with no dependencies, or on Cloudflare Workers and D1 behind a password, with signed, sliding sessions and a throttled login. The theme builder shows live contrast ratios and can nudge an accent until it passes.

Themes
63, checked by 819 automated contrast tests
Stack
Vanilla JavaScript, Node, Cloudflare Workers, D1
Code
6,130 lines
Built
September 2026
checklist-app.pages.dev
Checklist demo: the October calendar beside today's list, in the teal theme
The same day in the Midnight theme
The public demo, with sample tasks. Changes stay in your browser.

A last-one-standing Roblox arena where tiles break under your feet. Players dash, fire shockwaves and charge a quake ultimate across seven escalating levels, with ranked trophies and a global leaderboard.

The whole world is generated from code at runtime, and the server is authoritative, so the client cannot fake a hit or a coin.

Code
4,576 lines of Luau across 24 files
Tooling
Rojo, luau-analyze, Roblox DataStores
Built
June 2026
The core mechanic, drawn live in your browser.

A complete wedding planner in a single HTML file that works offline: budget tracker, guest list and RSVPs, drag-and-drop seating, vendor tracker, a twelve-month checklist and six themes. No install, no account, the data never leaves the couple's browser. I packaged it for Etsy.

A companion engine renders a 49-pin Pinterest library from HTML templates with headless Chrome and rotates it on a schedule with no repeats.

Planner
Vanilla JavaScript, fully offline
Pin engine
6 by hand · 1,704 lines of Python
Built
June 2026
Pin: where a $35,000 wedding budget goes
Pin: the wedding planner that does the math for you
Pin: the sage and champagne color theme
Three of the 49 pins the engine renders.

The commit ledger

Every commit across my repositories since April 14, 2026, read from GitHub and local branches. Ink squares are commits I wrote. A blue mark means an automation I built shipped something that day.

243commits by hand
371by automations
19repositories
42days with commits
May 27busiest day, 19 commits
LessMore Automation shipped Updated Oct 3, 2026

03

Experience

Sep 2026 – PresentJersey City, NJ

E*TRADE from Morgan Stanley

Business Analyst

  • Review paper and electronic account applications against “good order” standards, clearing deficiencies with submitting teams to prevent rework and resubmissions.
  • Process new account applications end to end, owning manual and exception cases to sustain daily volumes within service-level standards.
  • Reconcile daily exception reports against account records, tracing breaks to their source and escalating discrepancies to uphold risk controls.
  • Screen application data against internal risk criteria to identify potential fraud and high-risk accounts, escalating findings to compliance partners.
  • Resolve account inquiries from Operations and Customer Service, delivering documented outcomes within same-day turnaround targets.
  • Coordinate stock plan activations and account maintenance across internal teams, completing updates without service interruption.
Jun 2025 – Aug 2025New York, NY

S&P Global

Commercial Analyst

  • Built an RFP automation workflow in Python with an internal AI assistant, retrieving standardized answers from prior documents to populate new proposals; adopted across the team and cut turnaround from several hours to under one.
  • Analyzed financial agreement and prospectus data in internal models, presenting findings to four business units.
  • Consolidated weekly reporting for 15+ stakeholders, eliminating duplicate data pulls and manual formatting.
  • Ran testing and validation for a large client deal, performing financial data checks across large datasets.
Jun 2024 – Aug 2024New York, NY

AlphabloQ

Business Development & Commercial Analyst

  • Generated qualified leads through market research, marketing reports, and pipeline analysis, surfacing new business opportunities.
  • Built a client acquisition framework on customer relationship management (CRM) analytics while managing 30+ client accounts.
  • Translated complex financial and economic concepts into clear investor-facing content for LinkedIn and the company blog.
Jan 2024Baltimore, MD

HOPE Financial

Marketing & Outreach Intern

  • Ran homebuyer outreach across Maryland markets through targeted direct mail and segmented newsletters.
  • Applied data-driven marketing strategies to expand reach and improve email open and click-through performance.
Jul 2023 – Aug 2023San Diego, CA

Euler

Operations Intern

  • Evaluated partner relationship management (PRM) solutions and authored three operational reports identifying cost-saving opportunities across vendor workflows.
  • Rebuilt database operations and Excel-based customer data entry, reducing entry errors across two business lines.

04

Education & honors

Education

Williams College

Bachelor of Arts in Economics

Jun 2026 · Williamstown, MA

GPA 3.8Dean's List

University College Dublin

Study abroad, Finance and Economics (Arcadia Abroad)

Sep 2024 – Dec 2024 · Dublin, Ireland

Leadership & activities

Williams Economics Society

Vice President

Nov 2023 – Jun 2026

Ubuntu Advisory Group

Consultant

Jan 2024 – Jun 2026

Citadel Securities

Researcher / Analyst

Jul 2022 – Aug 2022

Scholarships & programs

Thrive Scholars

Scholar / Ambassador. Selected among 100 students nationwide for a six-year program supporting career development and networking.

Sep 2020 – Jun 2026

Houlihan Lokey

Corporate Scholar

Jul 2021 – Jun 2026

Hispanic Scholarship Fund

Scholar. Selected among 10,000 scholars from more than 100,000 applicants nationwide.

2024

Technical

Python · SQL · Advanced Microsoft Excel · Microsoft Power BI · Salesforce · Financial Modeling · Data Visualization

Languages

English (native) · Spanish (native)

Interests

Building automation projects · Business development · Behavioral economics · Entrepreneurship · Reading · Travel

05

Let's talk

Open to analyst, operations, automation and business development roles in New York or remote. Email is the fastest way to reach me, and I reply quickly.

aramis.jl04@gmail.com

Thanks for reading. Aramis