Are you a developer who wants to build this?

Praxiva started as developer research. We're inviting other developers to join, to build assistive AI that turns perception into real independence for people who need it.

How to join

Three simple steps

01

Reach out

Tell us a little about you and what draws you to this work. No formal application.

02

A short chat

We talk through your interests and where you might fit. Bring questions.

03

Start small

Take on a focused first project so we can build from there together.

Open source & open science

Code & experiment records

We're opening our source code and experiment documentation so others can inspect, reproduce, and build on the work, shared right here, in full.

praxiva-ai / research Public

Implementation code and experiment records for EnAct, Just Look, and Feel the Ball: assistive AI that turns perception into action.

Python License [ MIT / add ] Updated [ date ]
Files
  • experiments
  • models
  • enact.py
  • just_look.py
  • feel_the_ball.py
  • README.md
enact.py 42 lines · placeholder excerpt
1# enact.py: safety-aware reach guidance (excerpt)
2from models import VLMPlanner, DepthEstimator
3from haptics import Cue
4
5def guide_reach(frame, target):
6 """Guide a hand to `target`, checking the path each step."""
7 scene = VLMPlanner.ground(frame, target)
8 depth = DepthEstimator.run(frame)
9 if scene.collision_risk(depth) > 0.5:
10 return Cue.stop("obstacle ahead")
11 return Cue.toward(scene.next_step)
README.md · Experiment report

Praxiva Research

Code and experiment records behind our three ISEF systems. Everything needed to inspect, reproduce, and build on the work lives in this repository. [ Add a one-paragraph overview of the report. ]

Methods

Model architectures, training setup, and the reasoning pipeline for each system (enact.py, just_look.py, feel_the_ball.py). [ Summarize your methods. ]

Datasets & evaluation

  • Data sources and preprocessing: [ describe / link ]
  • Evaluation metrics and baselines: [ describe ]
  • Reproduction steps: [ commands to run ]

Results

[ Headline results and key tables/figures from the experiment report. ]

Code and report shown inline. Replace the placeholder excerpt and README text with your real content before launch.
Get in touch

Tell us you're interested

No commitment: just the first hello. We'll follow up and answer any questions.

Prefer email? Reach us at [ join@praxiva.ai ].