Create ProCogia organization card
Browse files
README.md
CHANGED
|
@@ -1,10 +1,78 @@
|
|
| 1 |
---
|
| 2 |
-
title:
|
| 3 |
-
emoji:
|
| 4 |
-
colorFrom:
|
| 5 |
colorTo: blue
|
| 6 |
sdk: static
|
| 7 |
pinned: false
|
| 8 |
---
|
| 9 |
|
| 10 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
+
title: ProCogia
|
| 3 |
+
emoji: 🧠
|
| 4 |
+
colorFrom: indigo
|
| 5 |
colorTo: blue
|
| 6 |
sdk: static
|
| 7 |
pinned: false
|
| 8 |
---
|
| 9 |
|
| 10 |
+
# ProCogia
|
| 11 |
+
|
| 12 |
+
**AI consulting, applied research, and enterprise AI products for regulated and data‑intensive industries.**
|
| 13 |
+
|
| 14 |
+
We build governable AI systems for pharma, health sciences, insurance/BFSI, manufacturing, and the public sector — combining domain‑adapted LLMs, execution‑parity validation, and private / on‑prem deployment.
|
| 15 |
+
|
| 16 |
+
🌐 [procogia.com](https://procogia.com) · 🧭 Vancouver, BC + North America · ✉️ [outreach@procogia.com](mailto:outreach@procogia.com)
|
| 17 |
+
|
| 18 |
+
---
|
| 19 |
+
|
| 20 |
+
## What we do
|
| 21 |
+
|
| 22 |
+
- **Enterprise AI products** — domain‑adapted LLMs and agentic systems built for regulated workflows
|
| 23 |
+
- **Applied AI consulting** — evaluation, observability, and governance for production LLM systems
|
| 24 |
+
- **Private AI infrastructure** — on‑prem and sovereign‑compute deployments with full audit trails
|
| 25 |
+
- **Custom model development** — supervised fine‑tuning and RLVR training with execution‑verified rewards
|
| 26 |
+
|
| 27 |
+
We ship AI that our clients can actually put in front of auditors, regulators, and customers.
|
| 28 |
+
|
| 29 |
+
---
|
| 30 |
+
|
| 31 |
+
## Product portfolio
|
| 32 |
+
|
| 33 |
+
### 🔀 PolyKode — SAS ↔ R ↔ Python code translation
|
| 34 |
+
Bidirectional code migration for regulated industries with **execution‑parity validation**: row/column checks, exact‑value parity, formatting preservation, and SAS/R behavioral‑difference handling. Built on a license‑aware corpus and a shared runtime that reproduces SAS‑exact numerical behavior (including MT19937 stream and `sgenrand` seeding).
|
| 35 |
+
|
| 36 |
+
- **Community edition:** [ProCogia/PolyKode-Community](https://huggingface.co/ProCogia/PolyKode-Community)
|
| 37 |
+
- **Enterprise edition:** Full parity harness, SAS 9.4 / Viya certification runs, audit trails, and regulated‑delivery SLAs. [Contact us →](mailto:outreach@procogia.com)
|
| 38 |
+
|
| 39 |
+
### 🧱 ZeroBoxx — private AI infrastructure
|
| 40 |
+
Turnkey on‑prem and sovereign‑cloud AI environments for organizations that cannot send data or model traffic to public inference APIs. Designed as the deployment surface for PolyKode Enterprise and other ProCogia products.
|
| 41 |
+
|
| 42 |
+
### 📞 CallYeah — voice AI for healthcare and services
|
| 43 |
+
Domain‑adapted voice AI for coverage, reminders, and scheduling workflows in healthcare and regulated services.
|
| 44 |
+
|
| 45 |
+
### 📊 Dataegis — governed data and analytics
|
| 46 |
+
Data platforms and analytics products for regulated industries, with governance and lineage built in.
|
| 47 |
+
|
| 48 |
+
---
|
| 49 |
+
|
| 50 |
+
## Research & training focus
|
| 51 |
+
|
| 52 |
+
Our current model development work centers on:
|
| 53 |
+
|
| 54 |
+
- **Execution‑verified reward signals** — RLVR with parseability, completion, semantic parity, robustness, and code‑quality rewards, replacing reference‑string similarity
|
| 55 |
+
- **Coverage‑driven training corpora** — 198‑construct SAS taxonomy, license‑segregated public and synthetic sources, and content‑addressed validation pipelines
|
| 56 |
+
- **Deterministic execution layers** — shared runtime routines that make SAS semantics reproducible across R and Python
|
| 57 |
+
- **Managed post‑training benchmarks** — template‑family‑disjoint holdouts and evaluation across prompting, SFT, and SFT+RL
|
| 58 |
+
|
| 59 |
+
We publish base checkpoints and community editions here. Fine‑tuned production models with parity harnesses and audit trails ship under Enterprise agreements.
|
| 60 |
+
|
| 61 |
+
---
|
| 62 |
+
|
| 63 |
+
## Partners & collaborators
|
| 64 |
+
|
| 65 |
+
ProCogia works with cloud and platform partners, systems integrators, government agencies, universities, and industry consortia across North America and Europe. We're actively building routes to market with partners in the SAS, Posit, and open‑source Python ecosystems, and evaluating multi‑partner consortia for sovereign‑compute and regulated‑industry AI programs.
|
| 66 |
+
|
| 67 |
+
---
|
| 68 |
+
|
| 69 |
+
## Get in touch
|
| 70 |
+
|
| 71 |
+
- 🌐 Website: [procogia.com](https://procogia.com)
|
| 72 |
+
- ✉️ General: [outreach@procogia.com](mailto:outreach@procogia.com)
|
| 73 |
+
- 💼 Partnership & enterprise: reach out via the website or email above
|
| 74 |
+
- 🧑💻 Careers: we hire across product, applied research, ML engineering, and consulting
|
| 75 |
+
|
| 76 |
+
---
|
| 77 |
+
|
| 78 |
+
<sub>ProCogia is headquartered in Vancouver, British Columbia. All models published under this organization are released with their upstream attribution and licenses preserved; see each model's card for details.</sub>
|