Parker Bidigare avatar

Parker Bidigare

Community

@sovr610 · Detroit Michigan

121Followers
|
281Public Repos
|
48Published Skills

Welcome! I have been a programmer all my life along side electronic design. Started with C++ and Java back when I was 12 years old. I do C#,python, js and more.

Agent Skills by Parker Bidigare

Showing 48 vetted skills indexed across 1 GitHub repositories.

sovr610sovr610

Paper-to-Spec Compiler

Fetch arXiv papers and compile them into machine-readable spec.yaml artifacts.

Community
Intermediate
sovr610sovr610

Fast Attention Path (PyTorch SDPA + optional FlashAttention-2)

Route PyTorch scaled dot product attention across Flash, Efficient, cuDNN, and Math backends.

Community
Advanced
sovr610sovr610

Meta-Learning Suite (MAML/FOMAML/Reptile + MAML++ Enhancements)

Execute meta-learning workflows for MAML, FOMAML, Reptile, and MAML++ enhancements.

Community
Advanced
sovr610sovr610

World Model Architecture Scaffold

Generate a modular world-model scaffold with ABC interfaces and Hydra-compatible YAML configs.

Community
Advanced
sovr610sovr610

V-JEPA 2 Video & Masking

Orchestrate V-JEPA 2 video tokenization and spatiotemporal masking for pretraining.

Community
Advanced
sovr610sovr610

Multi-Modal Alignment for Shared Embedding Space

Align encoder outputs from vision, text, audio, and sensor modalities into a shared embedding space.

Community
Advanced
sovr610sovr610

Neuro-Symbolic Engine (Fuzzy Logic + LTN/Real Logic Grounding)

Ground symbols to neural modules and evaluate fuzzy-logic rules over embeddings.

Community
Advanced
sovr610sovr610

Data Pipeline & Loaders

Register and consume phase-aware data loaders for brain_ai across 7 training phases.

Community
Advanced
sovr610sovr610

V-JEPA 2 Vision Transformer

Implement the V-JEPA 2 Vision Transformer architecture for image and video tasks.

Community
Advanced
sovr610sovr610

Hyperparameter Search

Automate phase-wise hyperparameter optimization for BrainAI training pipelines.

Community
Advanced
sovr610sovr610

Global Workspace Competition + Broadcast + Working Memory with Ignition Dynamics

Tokenize and fuse multi-modal encoder outputs into a constrained global workspace.

Community
Advanced
sovr610sovr610

Dual-Process Reasoning (System 1/2 + Metacognitive Control)

Route System 1 predictions and System 2 refinement with deterministic metacognitive control.

Community
Advanced
sovr610sovr610

BrainAI Spiking Core (LIF Family) + Surrogate Gradient Training

Implement explicit state contracts for LIF-family spiking neurons with surrogate gradients.

Community
Advanced
sovr610sovr610

Distributed Scaling

Configure DDP and FSDP distributed training for BrainAI models across GPUs and nodes.

Community
Advanced
sovr610sovr610

BrainAI System Orchestrator

Orchestrates BrainAI modules into deterministic pipelines with explicit contracts and bypass paths.

Community
Advanced
sovr610sovr610

Video DataLoader Pipeline (Decord + tf.data)

Build configurable video data pipelines with Decord PyTorch and tf.data TPU paths.

Community
Advanced
sovr610sovr610

Visualization & Interpretability

Render spike rasters, attention heatmaps, and reasoning traces from brain_ai internals.

Community
Advanced
sovr610sovr610

V-JEPA 2 Data Pipeline

Build and validate a scalable V-JEPA 2 video data pipeline with YAML-config-driven DataLoader construction.

Community
Advanced
sovr610sovr610

V-JEPA 2 Distributed Training

Distribute V-JEPA 2 training across GPUs and nodes with NCCL and SLURM.

Community
Advanced
sovr610sovr610

Evaluation & Benchmarks

Evaluate brain_ai models with standardized benchmarks and JSON/CSV reports.

Community
Advanced
sovr610sovr610

Active Inference Agent (Generative Model + EFE + Empowerment)

Build Active Inference agents that plan actions by minimizing Expected Free Energy.

Community
Advanced
sovr610sovr610

Task2Vec Task Embeddings + Curriculum/Clustering

Compute Fisher-based Task2Vec embeddings for episodic few-shot tasks and cluster them for curriculum workflows.

Community
Advanced
sovr610sovr610

Affective State Estimator (Valence-Arousal + Appraisal + Modulation)

Maps cognitive signals to Valence-Arousal-Dominance and outputs modulation targets for AI agents.

Community
Advanced
sovr610sovr610

DreamerV3-Style RSSM World Model

Build a recurrent state space model with BlockGRU and unimix sampling.

Community
Advanced