HomeAboutProjectsRadarSkillsAchievementsContact
🌙
All projects

SELECTED WORK / CASE STUDY

FINTECH / RESEARCH WORKFLOWS

StockBot.

Every recommendation deserves a paper trail.

StockBot brings scattered market recommendations into a private, traceable review workflow.

Visit project

  • Private research workspace

  • Source-to-decision tracking

BUILD IN PUBLIC / WORKFLOW FILM

The bit after “save for later”.

The original claim is only the start. This short film explains the research-intake idea: keep a source attached to the question, then separate the claim from the evidence needed to assess it.

An illustrative workflow from the existing StockBot media project, not a recording of a completed research run.

STOCKBOT / PRODUCT VIEW

StockBot workflow: source, evidence, and review
Product workflow illustration: preserve the source, check the evidence, and record the decision.

01 / THE BUILD

The starting point

A recommendation loses context when its source, timing, price levels, and follow-up live in separate places. StockBot makes those details available together so the user can review a claim and track what happened afterward.

02 / THE BUILD

The experience

  1. 01

    Capture a recommendation with its source, report date, entry, target, and stop.

  2. 02

    Review data-quality flags alongside price levels and risk context.

  3. 03

    Keep personal notes and decisions distinct from the source recommendation.

  4. 04

    Follow recommendations over time with outcome tracking and research views.

03 / THE BUILD

Under the hood

Next.js and TypeScript frontend on Vercel, backed by a Python FastAPI service and PostgreSQL on Railway. A server-side API proxy connects the interface to the backend, while Clerk controls access to the private workspace.

Keep the source separate from the decision

A recommendation is an input to research. Preserving its original wording and levels alongside personal notes makes it possible to revisit the decision without rewriting the source.

Make uncertainty visible

Missing fields, stale prices, and source mismatches are review signals. Surfacing them beside the research keeps gaps in the data from disappearing behind a polished summary.

Keep the workspace private

The public landing page explains the product. Approved access protects the research workspace and its user-specific records.

StockBot workflow overview
Workflow overview: the stages behind the product.

BUILT WITH

  • Next.js

  • TypeScript

  • FastAPI

  • PostgreSQL

  • Clerk

  • Playwright

  • Vercel

  • Railway

04 / THE BUILD

Making it dependable

Access controls separate the public landing page from private research and account data.

Data checks surface stale prices, missing fields, and source mismatches.

The interface preserves source context rather than presenting every recommendation as an action to take.

PROJECT STATUS

Public landing page is live. The research workspace requires approved access.

KEEP EXPLORING

Explore all projects
CrossTax