
3
3
Indonesia
1 year of experience
Hi, I'm Albert — a freelance full-stack developer. I design, build and ship custom software end-to-end: POS, CRM, payroll, internal tools and AI-powered workflows. One person, one point of accountability, real systems that run your business.

QuotePlot Agent is a full-stack market intelligence ecosystem that solves the core problem every financial AI hits: the data it needs is locked behind aggressive bot detection, JavaScript-rendered pages, and geo-blocks on financial sites like investing portals, market data providers, and financial news outlets. Bright Data's MCP server is the infrastructure layer that makes QuotePlot possible at production scale. The agent uses Bright Data's web_search and scrape_as_markdown tools to collect live price data, earnings announcements, and financial news from sources that would otherwise block automated access — feeding fresh, real-world signals into the analysis pipeline rather than relying on stale API snapshots. On top of that live data layer, QuotePlot runs a trained SVM pipeline that classifies user intent with 86% accuracy — distinguishing between price queries, trend analysis requests, and news sentiment questions — then routes each to the appropriate model. A predictive engine detects bullish and bearish trend signals from scraped historical and live price patterns, returning structured analysis through a FastAPI backend and visualised in real-time React charts. The result is a financial intelligence agent that operates on the actual live web — not cached data, not rate-limited public APIs — giving retail investors the same real-time market access that was previously reserved for institutional infrastructure.
31 May 2026

InsurancePortal is a production-deployed internal tool for life insurance agencies to store sold policies, track agent performance, and retrieve policyholder records instantly when customers walk in for inquiry. It is live at in.carlssonstudio.com and serves a real agency with real data. For this hackathon, IBM Bob was used as the primary development partner across four distinct feature implementations — each documented in the attached Bob report. First, Bob built a complete Claim model from scratch using Code mode: migration, Eloquent model with life insurance-specific claim types (death, maturity, surrender, critical illness, disability), ClaimController, FormRequest, ClaimObserver for auto-generating claim numbers, and the full Inertia React search interface — all in a single contextual session that respected existing codebase conventions without a single manual file directive. Second, Bob implemented an AI-powered dashboard using Plan then Code mode: aggregation queries across Policy, Claim, Nasabah, and Kwitansi models feeding six KPI cards and four Recharts visualisations, topped by an IBM Granite-generated narrative summary that reads agency performance in life insurance domain language — MDRT tracking, Empire Club progress, Gap Bonus eligibility — in both Bahasa Indonesia and English, toggled by the locale switch. Third, Bob ran in Orchestrator mode to sweep every React component and Inertia page, extract all hardcoded Indonesian strings, and generate both id.json and en.json translation files autonomously — the strongest repetitive automation showcase in the Bob report. Fourth, Bob generated a full PHPUnit and React Testing Library test suite across all new controllers and components in a single Orchestrator session — verifiable directly on the public GitHub repository.
17 May 2026

CommerceSystem is a comprehensive, production-ready enterprise solution designed to revolutionize modern retail through agentic workflows. The project is built on three robust pillars: the CommerceSystem-API, which serves as the central data hub; CommerceStore, a dynamic web storefront; and CommercePOS, a specialized point-of-sale interface for physical operations. By integrating Google Gemini via AI Studio, we transform these static components into an "Intelligent Enterprise." We deploy specialized AI agents across the ecosystem: a Sales Agent on the storefront to provide personalized recommendations and handle complex product queries, and an Operations Agent within the POS/API layer to monitor real-time stock levels, detect anomalies in sales, and automate replenishment orders. This multi-agent approach solves the common enterprise challenge of fragmented data between online and offline channels. Instead of manual data entry or rigid rules, our Gemini agents utilize long-context capabilities to analyze historical sales trends and customer interactions, providing actionable intelligence and automated task execution. This architecture reduces operational overhead and creates a seamless, intelligent bridge between digital and physical commerce environments.
19 May 2026