3.2 (3.2.5) PAL demos (openAI)
The PAL demos progressively build a Palantir-inspired data-intelligence platform, evolving from simple JSON analysis to cloud-hosted APIs, database-backed workflows, multi-stage analysis pipelines, advanced pattern detection, and local LLM deployment. These PAL demos all use openAI API except for 2.8 (uses Gemma). For details see
- Lab notes doc #603_PAL_.docx (PAL = Palantir) on the Gdrive.
TOC
- [2.1] pal v1 (BINGO) 26.0326. First working PAL prototype. Demonstrated basic ingestion, querying, and LLM-assisted analysis of structured event data using JSON files.
- [2.2] pal v2 (BINGO) 26.0327. Improved the PAL architecture and analysis workflow. Added more reliable data handling and expanded the deterministic processing pipeline.
- [2.3] pal v3 (BINGO) 26.0327. Refined the PAL concept into a more complete data-intelligence workflow. Focused on improving query, analysis, and reporting capabilities.
- [2.4] pal v4 (BINGO) 26.0328. Converted PAL into a FastAPI service with API endpoints. Demonstrated how PAL functionality could be exposed as a reusable service.
- [2.4b] pal_v4 deploy to render (BINGO) 26.0329-30. Deployed PAL v4 to the cloud using Render. Demonstrated remote access, API hosting, and basic production-style deployment.
- [2.5a/b] pal_v5 add Mongodb/deploy Render (BINGO) 26.0330. Added MongoDB persistence and cloud deployment. PAL moved from file-based storage to a database-backed architecture.
- [2.6] pal_v6 S0-S4 (BINGO) 26.0331-0401. Implemented a structured multi-stage analysis pipeline (S0-S4). Demonstrated how deterministic processing stages can be chained together for more complex workflows.
- [2.7] pal_v7 useful functions (BINGO) 26.0401. Added practical analysis functions such as entity history, recurring-event detection, rankings, comparisons, and pattern discovery. This version became the most useful operational PAL demo.
- [2.8] pal_v1 with local Gemma-4 (win11/gpu) 26.0411 (BINGO). Replaced the remote OpenAI model with a local Gemma-4 model running on Win11/GPU. Demonstrated that the PAL architecture is model-agnostic and can operate with local LLMs.
26.0618 (0520)