Full-time trader · Founder of PBCEx · Vibe-coding Next.js apps

• Now Trading / Now Building

• PBCEx – People’s Bank & Commodities Exchange

• Evidence-First History – Timeline & S.E.P.T.

• Other Builds

• AI Development Stack

• Product Stack

• Connect

I trade global markets and build rails for them at the same time.

• Trading: crypto, precious metals, select equities and indices

• Building: a tokenized-commodities exchange for real people, not just funds

• Goal: asset-backed rails where retail, producers, and institutions can all clear on the same stack

markets: crypto · commodities · equities

building: PBCEx · automation lab

location: Pittsburgh, PA

mode: shipping, not polishing

Right now I’m focused on:

• Shipping the PBCEx MVP (synthetic metals on an internal trading ledger w/ insured fulfillment)

• Price aggregation, hedging, and dashboards for metals and FX flows

• Agent-driven ops: ledgers, support drafts, monitoring, and journaling

PBCEx lets users buy, trade, spend, and redeem physical commodities.

• Early phase: synthetic metals (XAU-s, XAG-s, XPT-s, XPD-s, XCU-s) on an internal ledger

• Custody partners and vault-backed metals rather than pure paper exposure

• Physical fulfillment with insured logistics

• Spread + fee model targeting a lean, transparent market-of-record for metals

• Long-term plan: vault-backed tokens and a dedicated settlement stack (L1 + L2s)

Core public repos:

• abes-pbcex-workspace – monorepo for backend, frontend, and ops tooling

• PBCEx-Obsidian – knowledge base, specs, and architecture decision records (ADRs)

• Admin and ops terminals for risk, hedging, support, and investor tiles

I care about how we record events as much as how we trade them.

S.E.P.T. stands for: Source, Event, Place, Time.

Source S said that event E occurred at place P during time T.

The idea:

• Every entry has who, what, when, where, and why

• Evidence attached: people, places, things, photos, notes, papers, journals

• A chain of evidence: who wrote this, when, and who handled and submitted it?

• Meta tags:

  • People – e.g., “Martin Luther King Jr.”, “George Washington”

  • Places – e.g., “Shanghai, China”

  • Things – e.g., “AI”, “F-22 jet plane”

  • Lens – e.g., “Religion”, “Finance”, “Research”, “News”

We are usually reading the victor’s version of history. S.E.P.T. focuses on:

• Storyline = why and how, not just what

• Backlinks between correlated events

• Higher correlation and stronger evidence → higher probability and correlation scores

Repo:

• Timeline – evidence-based historical timeline with S.E.P.T. indexing

If you care about evidence-first history, and how we can tie this into automated analytics reach out and work with us on Source Timeline

Outside of PBCEx, I build tools where real-world operators need live data.

• Meat export & slaughterhouse dashboard – live beef prices and export data for a slaughterhouse builder/exporter, wired into a charting interface so they can track spreads and opportunities.

• Internal tools and experiments:

  • Micro-manufacturing and metals workflows

  • Ops dashboards for logistics and payments

  • Utilities around trading, journaling, and research

Most of these use the same core stack: Next.js, TypeScript, and AI-assisted workflows.

I run an AI-centered workflow anchored by a “second brain” system.

Context layer:

• Obsidian vault and GitHub repos (front-matter + tags) for mirrored indexs.

• Unified context across AI tools via retrieval-augmented generation (RAG) and model-context-protocol-style patterns

Workflow loop:

1. Plan and prompt engineering with GPT-class models using template/meta prompts

2. Build and test with Claude models

3. Use CodeRabbit / CodeX for pull-request gating and code review

4. Close each session with an ADR (architecture decision record) and session journal, then upsert to Pinecone

Quality control:

• Manual reviews after builds and again after test runs

• Final pass after PR reviews to avoid regressions and hallucinations

• Clear audit trail and fast iteration cycles

Terminal setup:

• iTerm2 + tmux running parallel command-line agents (Claude, Gemini, Grok, internal scripts)

• Repos wired for multi-agent workflows

Agents and use cases:

• Price aggregation agent – aggregates prices from several online sources to build a live range

• Customer-support drafting agent – drafts responses for support emails to a drafts folder for human review and send

• Markets agent – generates daily price charts and stats for hidden markets dashboards

Hybrid architecture

• Local models (NAS-backed) for sensitive data

• Cloud models for general reasoning

• Open WebUI Remix / open-source stacks as hybrid interfaces

• Running Qwen, LLaMA, and GPT-class models locally for experiments and “agent builder” projects

• Wrappers with Express / NestJS

Frontend

• Next.js / React

• Trading interfaces with embedded charting components

Backend

• Node.js / NestJS

• REST and streaming endpoints for ledgers, streaming dashboards, and price feeds

Data & Infra

• PostgreSQL (Supabase)

• Redis (Upstash)

• Docker, GitHub Actions CI/CD

• Cloudflare WAF/CDN

• Datadog and n8n for monitoring and automation

Markets & Payments

• Charts: TradingView (with advanced charting planned at launch)

• Payments: Stripe plus Plaid for merchant-of-record checkout

• Logistics: FedEx, Brink’s, Dillon Gage integrations

Creative Tools

• Adobe Illustrator (layered/path SVGs for react), Photoshop, Premiere

• Website: https://pbcex.com

• Source Timeline: – reach out and work with us on Source Timeline.

• LinkedIn: https://www.linkedin.com/in/ebraheim-salem-2117688b/

• X (Twitter): https://x.com/DesertFox_99

• Instagram: https://www.instagram.com/salem.ebraheim/