AI systems that remember
Durable AI memory
Memory should be scoped, inspectable, and useful on the next turn, not just more text stuffed into a prompt.
AI systems that remember, navigate, and prove themselves
I build the missing system layer around useful AI.
Memory that survives the next turn. Codebase context that starts in the right place. Evaluation and governance surfaces that make behavior inspectable before it matters.
Browse full project indexWhat I build
The work clusters into four practical capabilities a recruiter, founder, or technical lead can understand quickly. Project names are proof, not the headline.
AI systems that understand codebases
Repo-aware context broker
Coding agents need a map of the repository: what matters, what changed, what is adjacent, and when to stop widening.
AI systems that can be evaluated and governed
Behavioral eval harness
Useful AI systems need proof surfaces: evals, verdicts, promotion gates, and failure modes that operators can understand.
Behavioral eval harness Proof: Fenrir Condition-bounded LLM behavior tests with canonical readouts and uncertainty guardrails. Access-governance contracts Proof: Heimdall Typed evidence goes in; bounded verdicts and downstream response contracts come out. Capability promotion gates Proof: Sindri Verifier-backed review bundles and graduation criteria for proposed automation capabilities.
Applied workflow demos
Clinical documentation demo
Applied demos show how the system layer feels in a real workflow: reviewable, bounded, and built around human sign-off.
Built from the operator side
I am a systems and applied AI engineer focused on the parts that decide whether AI becomes useful in production: context, memory, routing, evaluation, and workflow fit.
The standard is practical usefulness under real constraints, not a demo that only works once.
Where I fit
If you are building an AI product, evaluating an agent, modernizing an internal workflow, or trying to make a code assistant behave reliably inside a real repository, I work on the control layer around the model.
That means designing the task boundary, the memory contract, the repo context path, the governance surface, and the evidence that tells you whether the system is improving.
How I think about AI systemsFull project index
The complete index includes core systems, experimental lanes, applied demos, and commissioned competency work. It is here for depth after the quick capability scan.
Core
Bifrost
Systems Builder Core Access on request
Cross-project atlas that defines ownership boundaries, capability contracts, terminology, and overlap-risk closure across the ecosystem.
Echo
Applied AI Engineer Core Access on request
AI-native clinical documentation demo that turns simulated encounters into reviewable notes with clinician sign-off.
friday
Systems Builder Core
Assistant orchestration proving ground where routed lanes, memory providers, and failure handling are exercised under runtime constraints.
Heimdall
Systems Builder Core Access on request
Bounded adjudication and access-governance service for assistant verdicts and response contracts.
LAILA
Applied AI Engineer Core Access on request
Next-generation assistant runtime unifying agency, Muninn memory, and identity continuity.
lex
Applied AI Engineer Core Access on request
Client-facing local-first companion runtime centered on identity continuity, persona control, and avatar-aware interaction.
Mimir
Applied AI Engineer Core Access on request
Repository self-model and world-state orientation system that brokers bounded rehydration for coding agents across evolving codebases.
muninn
Applied AI Engineer Core
Model-agnostic memory substrate with card, evidence, and policy-state primitives plus deterministic rehydration for agent runtimes.
OnHand
Applied AI Engineer Core Access on request
Inventory and intent platform spanning Android client workflows, enterprise APIs, and optional vision evaluation services.
PulseTrade
Applied AI Engineer Core
Commissioned applied competency project for modular trading research and policy-driven execution planning.
Experimental
Althing
Systems Builder Experimental Access on request
Profile-aware specialist-routing harness for multi-lane inference experiments with explicit degraded modes and measurable control profiles.
Fenrir
Applied AI Engineer Experimental
Hybrid behavioral evaluation tool for condition-bounded LLM behavior under pressure.
Gauntlet
Automation + Quality Experimental
Coding arena platform for challenge publishing, controlled execution, and repeatable submission scoring.
MemeTrader
Applied AI Engineer Experimental
Commissioned applied competency project for deterministic strategy evaluation and replayable market decisions.
Null_Signal
Applied AI Engineer Experimental
Local-first autonomous media pipeline that converts news signals into scripted, rendered, and packaged short-form broadcasts.
ReadyPlayer1
Applied AI Engineer Experimental
Automated playtesting system built specifically to exercise SubSim through deterministic simulation runs.
Sindri
Applied AI Engineer Experimental Access on request
Bounded capability-forging service focused on trust-calibrated wrapper lanes, verifier-backed review bundles, and promotion gating.
SNN
Applied AI Engineer Experimental Access on request
Neuromorphic research sandbox for spiking networks, memory-core experiments, and SpikingBrain integration.
subsim
Developer Experience Experimental
Audio-first deterministic submarine simulation designed for fast iteration, replay traces, and headless testing.
Commission
Accountant
Systems Builder Commission Access on request
Commissioned applied competency project for secure accounting ingestion, OCR review, and export workflows.
Architecture map
For readers who want the topology, this graph shows how the systems relate beneath the capability buckets.
-
Assistant orchestration proving ground where routed lanes, memory providers, and failure handling are exercised under runtime constraints.
Client and identity surface
-
Client-facing local-first companion runtime centered on identity continuity, persona control, and avatar-aware interaction.
Neuromorphic memory research
-
Neuromorphic research sandbox for spiking networks, memory-core experiments, and SpikingBrain integration.
-
Routing and lane instrumentation
-
Profile-aware specialist routing with constrained research lanes and non-authoritative shadow characterization.
Profile-aware specialist-routing harness for multi-lane inference experiments with explicit degraded modes and measurable control profiles.
-
Adjudicates typed evidence into bounded verdicts and downstream assistant response contracts.
Bounded adjudication and access-governance service for assistant verdicts and response contracts.
Applied product demos
-
AI-native clinical documentation demo that turns simulated encounters into reviewable notes with clinician sign-off.
Behavior evaluation
-
Tests condition-bounded LLM behavior shifts under pressure with explicit uncertainty guardrails.
Hybrid behavioral evaluation tool for condition-bounded LLM behavior under pressure.
Experimentation waves
-
Audio-first deterministic submarine simulation designed for fast iteration, replay traces, and headless testing.
Automated playtesting
-
Automated playtesting system built specifically to exercise SubSim through deterministic simulation runs.
-
-
Local-first autonomous media pipeline that converts news signals into scripted, rendered, and packaged short-form broadcasts.
-
Coding arena platform for challenge publishing, controlled execution, and repeatable submission scoring.
Commission work (applied competencies)
-
Commissioned applied competency project for secure accounting ingestion, OCR review, and export workflows.
Inventory/vision exploration
-
Inventory and intent platform spanning Android client workflows, enterprise APIs, and optional vision evaluation services.
-
-
Commissioned applied competency project for modular trading research and policy-driven execution planning.
-
Commissioned applied competency project for deterministic strategy evaluation and replayable market decisions.
-
-
Started as embedded memory experiments inside FRIDAY/Lex.
Model-agnostic memory substrate with card, evidence, and policy-state primitives plus deterministic rehydration for agent runtimes.
Now a standalone memory substrate for cards, evidence, policy-state, and deterministic rehydration.
Repository orientation branch
-
Extends memory work into repository world-state orientation, brokered context, and bounded cross-repo export.
Repository self-model and world-state orientation layer with deterministic brokered rehydration and cross-repo validation across real engineering repos.
Ecosystem atlas and boundary governance
-
Cross-project atlas that defines ownership boundaries, capability contracts, terminology, and overlap-risk closure across the ecosystem.
Bounded capability proving lanes
-
Wrapper lane graduated; adjacent-lane transfer proving remains intentionally narrow and safety-gated.
Bounded capability-forging service focused on trust-calibrated wrapper lanes, verifier-backed review bundles, and promotion gating.
-
-
-
Next-generation assistant runtime unifying agency, Muninn memory, and identity continuity.
Interested in working together?
Whether you want help designing a trustworthy AI system layer, pressure-testing an architecture direction, or exploring Mímir as a foundational component, I am open to serious conversations.