Quantura: Building an Inspectable Forecasting Platform
1
0
Scout: AI-assisted forecast research — Quantura
I built Scout inside Quantura to help users interrogate the history and forecast they are actually viewing. Three contextual question cards organize the experience around Forecast, History and Evidence.
Users can inspect median changes and uncertainty ranges, ask about observed highs and lows, and review data sources and units. Structured answers use the selected observations and forecast values as context.
The interface also supports chart annotations, user notes, saved conversations, and named CSV previews before forecasting. The goal is to make a forecast easier to inspect, explain and revisit.
My work combines TypeScript backend logic, a JavaScript research interface and interactive charting.
Explore: https://quantura.studio/forecasting
0
9
Forecast API, SDKs & OAuth MCP — Quantura
I built a shared Search → Download → Forecast API for Quantura so developers can move from a discovered market or dataset to historical observations and asynchronous forecast results.
The integration includes JSON/CSV exports, job status handling, provider and model metadata, scoped API access, and OAuth sign-in for MCP clients. Python and TypeScript SDKs plus a CLI provide reusable integration examples.
I also organized the public RapidAPI listing into clear endpoint groups, documented usage and authentication, and added examples for scheduled forecasting workflows.
This work is useful for teams connecting forecasts to their own apps, notebooks or AI tools.
Explore the API: https://rapidapi.com/tamzid2001/api/quantura2
0
15
Forecasting pipelines across market and economic data
Built as part of Quantura, my forecasting and research product. I implemented a shared data workflow for Alpaca, Dukascopy, Gemini, Treasury Fiscal Data, World Bank Data360 and BigQuery public datasets.
The work covers provider discovery, series selection, timestamp and unit normalization, observed-history downloads, six forecast quantiles and scheduled execution. Data coverage and model requirements remain explicit, and missing observations are not invented.
Deliverables include Python/TypeScript integrations, JSON/CSV exports, asynchronous job handling and chronological historical evaluation.
I can build a similar workflow for your supported time-series data, with clear inputs, milestones and operating costs.
Explore: https://quantura.studio/
0
19