Projects as Systems

Technical case studies across runtimes, memory, protocols, retrieval, workflows, and AI infrastructure.

# active

emotional-memory

Research-grade memory layer for LLM systems, with affective state encoding and reproducible benchmarks against Mem0, LangMem, and Letta.

#Python#LLM Memory#Research#PyPI#DOI

Problem

LLM memory systems often claim persistence without reproducible evidence for recall quality, compression behavior, or state evolution.

System Design

A research-grade memory layer with affective state encoding, benchmark artifacts, and comparison surfaces against existing memory frameworks.

Architecture

  • memory store
  • affective encoding
  • benchmark runner
  • claim matrix

Runtime Model

  • ingest
  • encode
  • retrieve
  • evaluate
  • publish artifacts

Tooling

  • Python
  • PyTorch
  • pydantic
  • pytest
  • Zenodo

Reliability

  • reproducible benchmark runs
  • versioned claims
  • test-backed package release
ingestencoderetrieveevaluatepublish artifacts

Constraints

The public language must stay aligned with evidence. Stronger scientific claims require broader external validation.

Tradeoffs

Research rigor is prioritized over broad framework compatibility or feature surface.

Future

Expand human evaluation, semantic confound tests, and longitudinal memory benchmarks.

# active

orka

Agent runtime for routing work from chat and HTTP channels into prioritized LLM workflows with MCP/A2A support.

#Rust#MCP#A2A#RAG#WASM

Problem

AI work arriving from chat and HTTP channels needs routing, prioritization, tool access, and runtime boundaries instead of ad hoc automation.

System Design

An agent runtime for prioritized LLM workflows with MCP/A2A support and channel adapters.

Architecture

  • channel ingress
  • runtime scheduler
  • tool layer
  • MCP/A2A adapters

Runtime Model

  • receive
  • classify
  • route
  • execute
  • checkpoint

Tooling

  • Rust
  • tokio
  • WASM
  • MCP
  • A2A

Reliability

  • typed runtime boundaries
  • prioritized execution
  • protocol separation
receiveclassifyrouteexecutecheckpoint

Constraints

The runtime must stay protocol-oriented and avoid coupling to a single interface.

Tradeoffs

A lower-level runtime gives stronger control but requires sharper product boundaries.

Future

Harden observability, workspace policy, and durable execution semantics.

# active

msg2agent

Secure message transport for agent systems, combining end-to-end encryption, DID identity, relay delivery and protocol adapters.

#Go#E2E Encrypted#DID#A2A#MCP

Problem

Agents need to communicate across organizational boundaries without shared keys, central trust, or plaintext relay access.

System Design

A trustless agent-to-agent communication layer using W3C DID identity, end-to-end encryption, MCP connector support, and A2A interoperability.

Architecture

  • DID identity
  • encrypted message envelope
  • relay
  • MCP connector
  • A2A adapter

Runtime Model

  • discover agent
  • encrypt
  • sign
  • relay
  • pull inbox
  • acknowledge

Tooling

  • Go
  • X25519
  • Ed25519
  • W3C DID
  • OAuth 2.1 + PKCE
  • MCP

Reliability

  • store-and-forward delivery
  • offline inbox
  • tenant quotas
  • relay cannot read plaintext
discover agentencryptsignrelaypull inboxacknowledge

Constraints

The relay routes messages but must not become the trust anchor for identity or message confidentiality.

Tradeoffs

Cryptographic ownership increases trust clarity while adding connector and key-management complexity.

Future

Broaden connector distribution, production billing, and interoperability surfaces.

# active

OpenFatture

Italian invoicing system that combines FatturaPA/SDI compliance, Lightning payments and local-first AI assistance.

#Python#Lightning#Ollama#FatturaPA#CLI

Problem

Italian invoicing workflows combine compliance, payment rails, and assistance needs that do not fit generic SaaS automation.

System Design

A local-first invoicing system with FatturaPA/SDI compliance, Lightning payments, and controlled AI assistance.

Architecture

  • invoice model
  • SDI export
  • payment rail
  • local AI assistant

Runtime Model

  • draft
  • validate
  • export
  • send
  • reconcile

Tooling

  • Python
  • Streamlit
  • LND
  • Ollama
  • OpenAI/Anthropic

Reliability

  • local-first control
  • compliance validation
  • manual review before external effects
draftvalidateexportsendreconcile

Constraints

Compliance and accounting correctness are higher priority than autonomous action.

Tradeoffs

Local-first operation limits convenience but improves control over sensitive business data.

Future

Expand workflow checks, reconciliation, and auditable AI assistance.