Home › Tracks › Agentic Systems & Harness Engineering
Elite track
Orchestrate deterministic fleets of cloud-native agents with rigorous outer-loop verification.
Master advanced harness engineering: outer-loop verification systems, Just-In-Time (JIT) Context Hubs with semantic routing, cloud-native sandboxed code execution, and strict human-in-the-loop (HITL) guardrails. You build enterprise-grade agent platforms optimized for 2026 model economics (Luna, Sol, Muse Spark), bridging directly to Operating Systems, Distributed Systems, and Verification Compilers.
Week by week
Every week unlocks the next. Concepts route you to free, world-class material; projects turn that knowledge into something deployed.
Week 1
Engineer deterministic execution harnesses, multi-turn agent run-loops, and structured context state management separating core reasoning from environment orchestration.
Bridges to Operating Systems — processes, scheduling, and the run loopBuilds on: nothing, start here
Week 2
Architect agent tool harnesses using mid-session dynamic tool addition without KV-cache invalidation. Transition away from rebuilding static tool arrays every turn to incremental Model Context Protocol (MCP) bindings, granular cache breakpoints, and schema management.
Bridges to Software Engineering — interface design and API contractsBuilds on: Foundations of Harness Engineering & Loop Crafting
Week 3
Implement the Model Context Protocol (MCP) to standardize tool interfaces, contextual resource exposure, and dynamic prompt templates, moving beyond context-bloating raw schemas to sandboxed code execution as tools.
Bridges to Computer Networks — protocols, client-server architecture, and RPCBuilds on: Dynamic Inline Tooling, MCP & Cache-Preserving Toolmaking
Week 4
Engineer harnesses for always-on adaptive reasoning models where internal reasoning tokens cannot be toggled off. Scale test-time compute via search, dynamic thinking token allocation, refusal category handling, and outer-loop verification.
Bridges to Artificial Intelligence — search, planning, and state-space reasoningBuilds on: Dynamic Inline Tooling, MCP & Cache-Preserving Toolmaking
Week 5
Engineer low-latency JIT context delivery and dynamic routing using sub-10ms single-forward-pass semantic routers, hierarchical memory tiers, and cache-preserving vector retrieval.
Bridges to Operating Systems — memory hierarchy, paging, and cachingBuilds on: Planning, Adaptive Reasoning Scaling & Test-Time Compute
Week 6
Implement defense-in-depth agent containment using hypervisor MicroVMs (Firecracker), application kernel sandboxes (gVisor), eBPF execution tracing, and out-of-band watchdog monitoring to enforce strict non-bypassable runtime isolation for autonomous agent tools.
Bridges to Operating Systems — virtualization, namespaces, and process isolationBuilds on: Context Hubs, JIT Context Delivery & Semantic Routers
Week 7
Design and build a self-organizing multi-agent research fleet that dynamically negotiates specialist roles, executes reflective intermediate audits, and utilizes dynamic inline MCP tool calls without KV-cache invalidation.
Why it matters: Enterprise harness engineering for self-organizing agent fleets on frontier reasoning models.
A working self-organizing agent framework with dynamic role negotiation, prefix-cached MCP tool bindings, and an isolated microVM verification harness.
Week 8
Replace brittle static DAG multi-agent chains with self-organizing agent teams (SAT). Architect systems with dynamic role assignment, conversational reflection phases, and independent corrective step audits that beat static routing pipelines.
Bridges to Distributed Systems — coordination, message passing, and consensusBuilds on: Hardware-Isolated Sandboxing, Out-of-Band Watchdogs & Kernel Confinement
Week 9
Transition from manual syntax production to high-level intent specification, architectural task decomposition, and autonomous verification harness loops that compile and rigorously test agent-generated code.
Bridges to Compilers — program optimization, intermediate representations, and iterative refinement loopsBuilds on: Planning, Adaptive Reasoning Scaling & Test-Time Compute
Week 10
Architect strict Human-in-the-Loop (HITL) guardrails and automated verification outer loops. Replace fragile 'absolute autonomy' with deterministic testing harnesses that validate code-review safety and runtime constraints before commit.
Bridges to Software Engineering — testing, regression suites, and observabilityBuilds on: Self-Organizing Agent Teams, Dynamic Role Negotiation & Step Auditing
Week 11
Harden autonomous agent runtimes against indirect prompt injection, unvetted tool execution, and privilege escalation through credential scoping, kernel-isolated sandboxing, and out-of-band monitoring.
Bridges to Information Security — threat modeling, capability isolation, and least privilegeBuilds on: Agent Verification Loops & HITL Guardrails
Week 12
Design a highly resilient, cloud-native asynchronous execution harness. Build an inner loop that generates and modifies complex repositories inside stateful sandboxes, and an outer loop that validates compilation, runs tests, and applies strict Human-in-the-Loop (HITL) gates before committing changes.
A complete, production-grade asynchronous agent platform with remote container sandboxes, a semantic-routing JIT Context Hub, and an automated verification harness with a web-based approval interface.
What's next
Each track stands alone, so there's no wrong order. If you want a suggestion, this one pairs well next.