Product / UX
Product scope, workflows, responsive UI, dark/light experience and multilingual foundations.
CASE STUDY / 01 · SOLO-BUILT PRODUCT
AurumTrade
AT/01PRODUCT SYSTEM
Zbudowany samodzielnie end-to-end przez Denysa Novikova: full-stackowy produkt AI-assisted do analizy tradingu, łączący trade capture, dane z giełd, behavioral journal, playbook discipline, analytics i subscription infrastructure.
Founder · Product EngineerFull-Stack · AI · DataSolo-built end-to-end
21product views
5AI workflows
17CSV variants
27RLS policies
01 / CAPTUREManual + CSVTrades, screenshots, context
02 / INTERPRETAnalyticsP&L, R:R, drawdown, behavior
03 / REFLECTJournal + PlaybookEmotion, rules, discipline
04 / ASSISTPersonalized AIReview, coach, psychology, chat
01 / PRODUCT PROBLEM
Dane tradingowe są rozproszone.
Użyteczny trading journal nie może kończyć się na tabeli zysków i strat. Produkt łączy structured trades z eksportami giełdowymi, screenshotami, jakością wykonania, emocjami, journal notes, zasadami playbooka i analizą retrospektywną — a następnie udostępnia ten kontekst analytics i AI-assisted review.
02 / SOLO OWNERSHIP
One product.
End-to-end ownership.
Full-Stack
21 application views, 15 API route modules, authentication, CRUD, imports, uploads and account lifecycle.
Data / PostgreSQL
Supabase schema, RLS ownership, indexes, triggers, constraints and database-level free-tier enforcement.
AI Systems
Five authenticated OpenAI workflows grounded in trades, playbooks, journal entries and emotion data.
Analytics
Performance metrics, equity curve, behavioral feedback and a 200-path Monte Carlo-style projection.
SaaS / Billing
Stripe Checkout, Customer Portal, signed webhooks and server-side AI entitlement foundations.
03 / SYSTEM ARCHITECTURE
Data in. Context out.
INPUT / 01Manual TradeTrade context + execution notes
INPUT / 02Exchange CSVDetection + mapping + normalization
→PRODUCT CORENext.jsUI + authenticated APIs
→DATASupabaseAuth · PostgreSQL · RLS · Storage
→OUTPUTAnalytics + AIPerformance · journal · coaching
Stripe Checkout→Signed Webhook→Plan State→AI Entitlement
04 / DATA INGESTION
17 export variants.
One ingestion path.
01Detect
→02Preview
→03Map
→04Normalize
→05Insert
Wybrane formaty eksportów giełdowych plus manual mapping. Nie jest to przedstawiane jako universal lub lossless exchange reconciliation.
05 / AI CONTEXT ARCHITECTURE
Personal history,
not generic commentary.
TRADESJOURNALPLAYBOOKEMOTIONS
CONTEXT ASSEMBLYOpenAI / GPT-4oAuthenticated · server-side · plan-gated
01Trade Review
02Trade Score
03Coach
04Psychology
05Chat
AurumTrade nie deklaruje przewidywania market outcomes. Warstwa AI to personalized review i coaching oparty na własnych danych użytkownika.
06 / DATA + ACCESS
Access rules live
below the interface.
8application tables
27RLS policies
18indexes
20free-plan trade cap
07 / ANALYTICS
Performance + behavior,
one feedback loop.
Win rateTotal / avg P&LR:RDrawdownLong / shortBest setupStreaksEquity curvePair breakdownWeekday / monthGrade / emotionRecent trades
200 paths
Interactive Monte Carlo-style performance projection oparty na historical win rate i average win/loss values. Simplified projection, nie market simulator ani predictive model.
08 / PRODUCT MATURITY
Complete beta product.
Known hardening work.
Forensic audit potwierdził szeroką implementację end-to-end i wskazał kolejny etap engineeringu: schema reconciliation, mocniejszą trade-domain validation, AI contracts, billing lifecycle hardening, privacy alignment, automated tests, CI/CD i operational observability.
09 / OUTCOME
Od rozproszonych trading artifacts do jednego connected product system.
AurumTrade pokazuje solo end-to-end product engineering obejmujący UX, full-stack architecture, data security, analytics, AI-assisted workflows i SaaS foundations — bez sprowadzania pracy do dashboardu.