You will get a custom local MCP AI agent for financial anomaly detection
Project details
You will get a private, working AI system that scans your assets for anomalies and lets you query it in plain language through an MCP-connected agent — no data has to leave your own machine or cloud environment. I'm a Senior Software Architect with 15+ years building high-performance financial systems, including work inside Brazil's B3 exchange ecosystem, and this exact stack has already caught a real 57% six-month decline in a NASDAQ stock purely from reconstruction-error patterns — before I'd even looked at the price chart myself. You're not buying a prototype: it's a PyTorch autoencoder, Qdrant vector search, and a local LLM via Ollama, running on consumer-grade hardware for under $1,200 in total cost. No recurring API bills, no vendor lock-in, and full source-code ownership available on the top tier.
Machine Learning Tools
Python, PyTorchWhat's included
| Service Tiers |
Starter
$300
|
Standard
$900
|
Advanced
$2,750
|
|---|---|---|---|
| Delivery Time | 3 days | 7 days | 14 days |
Number of Revisions | 1 | 2 | 2 |
Number of Model Variations | 1 | 2 | 3 |
Number of Scenarios | 5 | 15 | 30 |
Number of Graphs/Charts | 3 | 6 | 10 |
Model Validation/Testing | - | ||
Model Documentation | - | ||
Data Source Connectivity | - | ||
Source Code | - | - |
Optional add-ons
You can add these on the next page.
Fast Delivery
+$75 - $300
Additional Revision
+$50
Additional Model Variation
(+ 2 Days)
+$75
Additional Scenario
(+ 1 Day)
+$25
Additional Graph/Chart
+$15
Model Validation/Testing
(+ 2 Days)
+$150
Model Documentation
(+ 1 Day)
+$100
Data Source Connectivity
(+ 3 Days)
+$200
Source Code
(+ 2 Days)
+$650Frequently asked questions
About Rafael
AI & Machine Learning | Software Architecture
Blumenau, Brazil - 10:06 pm local time
Steps for completing your project
After purchasing the project, send requirements so Rafael can start the project.
Delivery time starts when Rafael receives requirements from you.
Rafael works on your project following the steps below.
Revisions may occur after the delivery date.
Confirm your requirements
I review your submitted tickers/assets, data source, analysis time period, and hosting preference (on-premise vs. cloud) before writing any code, so the build matches what you actually need from day one.
Run the anomaly detection model
I run the PyTorch autoencoder against your specified assets, generate the reconstruction-error analysis, and produce the report and charts included in your tier.