The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Back to feedPost
I'm Stanley, a full-stack dev who ships solo. Recently launched PetEats (AI iOS app, on the App Store) and Net Pay (netpaytoday.com). I build AI chatbots, automation, Telegram bots and full-stack apps — Next.js, PHP, Python, OpenAI. Open for freelance work. DMs open
netpaytoday.com
Net Pay — Salary, Loan & Savings Calculators (UK, US, AU, DE, IE)
Free take-home pay calculators for the UK, US, Australia, Germany and Ireland, plus loan and savings tools. Instant, accurate, private — nothing leaves your device.
The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Trending
retrofuturism
Share your favorite designs from the past exploring the distant future. Or your own take on the aesthetic 🚀
halftone
Dots. Texture. Grain. Let's see your designs using the halftone technique.
Midjourney
Creatives are showing off what they're making with Midjourney.
colorgrade
Video editors sound off: How do you enhance the mood of your video in color grading?
Related posts
The hardest part wasn't building the game. It was making it PLAYABLE. 🎮
While building Shadow Heist, I hit errors that broke the game, interrupted missions, and forced me back into debugging more times than I expected.
But that's where the real development happened.
I fixed the issues, tested the gameplay, pushed through the final mission, and turned the idea into something people can actually play.
Build. Break. Debug. Fix. Repeat.
That's what development really looks like behind the screen.
What's the toughest bug you've ever had to solve? 👇
#WebDevelopment #GameDevelopment #JavaScript #FullStackDevelopment #BuildInPublic
Toughest for me was a rich text editor that swallowed the first keystroke when it took focus. No error, no crash, the text just published with the first letter missing. It stayed hidden a while precisely because nothing failed loudly. Now I read the field back before submitting....
Rollout — a real-time 3D music launch platform I built.
An artist drops one finished song and gets a whole release back: AI cover art on a layered canvas, a beat-synced lyric video, a release calendar with pre-written captions, a hosted fan page, and an ad campaign. One platform, start to finish.
The landing is hand-written WebGL and GLSL. A raymarched glass blob that morphs and reacts to your cursor sits behind a real 3D headphones model, with drifting smoke and a star field, all running in a single WebGL context so it holds up across devices instead of fighting for the GPU.
Under the hood it's a real product, not a demo: multi-tenant SaaS on Supabase with Postgres row-level security, Stripe billing with server-enforced plans, and a Python engine that reads the song's key, tempo and mood and turns that reading into the whole visual and marketing direction.
Stack: vanilla JS + WebGL, Supabase, Stripe, Netlify, and a CLAP audio model for the vibe read.
love the shade of violet you selected, super definitive in & of itself! add in those sleek, clear 3d elements, & you've undoubtedly made a memorable impression on your user. fantastic!
Agentic Financial Document Intelligence & Transaction Reconciliation Engine
• Role: Lead AI & Systems Architect
• Deliverables: Multi-Modal PDF Parser, Hybrid Agent Orchestrator, Deterministic Math Engine, Audit Trail Logger
• Timeline: 3 Weeks
OVERVIEW
Financial institutions and back-office teams spend thousands of manual hours reconciling unstructured documents (loan contracts, invoices, compliance bylaws) with relational transaction ledgers. Standard LLM wrappers fail catastrophically in this domain due to mathematical hallucinations, version conflicts, and a lack of explainable audit trails.
I engineered an agentic document intelligence system that combines multi-modal PDF extraction with deterministic validation tools to automate complex cross-referencing with 100% computational precision.
THE PROBLEM & DIAGNOSIS
LLM Math Hallucinations: Generative models cannot be trusted to perform exact financial calculations, interest compounding, or tax audits.
Unstructured vs. Tabular Asymmetry: Document facts are scattered across messy layouts, whereas transaction logs live in strict relational schemas.
Document Version Discrepancies: Clauses change across revisions; systems must resolve conflicting policies before executing reconciliation.
KEY ARCHITECTURAL DECISIONS
• Decision 01: Separation of Reasoning and Execution
The LLM acts strictly as a reasoning and extraction layer, identifying entities and formulas. All financial calculations and reconciliations are handed off to deterministic Python execution sandboxes.
• Decision 02: Cross-Source Entity Disambiguation
Built a cross-referencing engine that maps unstructured text snippets directly to SQL transaction IDs using hybrid vector-keyword scoring.
• Decision 03: Fully Auditable Chain-of-Thought
Every validation output includes a step-by-step citation trail pointing to the exact page, bounding box, and ledger entry.
MEASURABLE RESULTS & IMPACT
• 0% computational hallucination rate achieved via deterministic code execution.
• 90%+ precision in entity extraction across multi-page, complex financial agreements.
• Production-grade audit trails satisfying compliance requirements.
TECH STACK
Python 3.11, FastAPI, Pydantic, LangChain, PyMuPDF, Pandas, PostgreSQL / pgvector, Docker.
