Looks like you have JavaScript disabled. For the full Twine experience, you will need to re-enable it.

Self-taught AI systems builder in Montreal focused on long-running LLM agents, multi-agent governance, drift detection, evidence routing, and agent workflow QA. I specialize in finding system failures, agent loops, routing errors, hallucinated claims, broken assumptions, missing evidence, and UI/agent behavior problems. I build and evaluate durable multi-agent ecosystems with provenance and safety gates, last-known-good restore workflows, and structured research outputs (logs, evidence surfaces, experimental plans) where written proof of work matters more than formal credentials.…Self-taught AI systems builder in Montreal focused on long-running LLM agents, multi-agent governance, drift detection, evidence routing, and agent workflow QA. I specialize in finding system failures, agent loops, routing errors, hallucinated claims, broken assumptions, missing evidence, and UI/agent behavior problems. I build and evaluate durable multi-agent ecosystems with provenance and safety gates, last-known-good restore workflows, and structured research outputs (logs, evidence surfaces, experimental plans) where written proof of work matters more than formal credentials.

Sean David Ramsingh

QA Engineer, AI Engineer, AI Strategy Consultant, +4





Self-taught AI systems builder in Montreal focused on long-running LLM agents, multi-agent governance, drift detection, evidence routing, and agent workflow QA. I specialize in finding system failures, agent loops, routing errors, hallucinated claims, broken assumptions, missing evidence, and UI/agent behavior problems. I build and evaluate durable multi-agent ecosystems with provenance and safety gates, last-known-good restore workflows, and structured research outputs (logs, evidence surfaces, experimental plans) where written proof of work matters more than formal credentials.…Self-taught AI systems builder in Montreal focused on long-running LLM agents, multi-agent governance, drift detection, evidence routing, and agent workflow QA. I specialize in finding system failures, agent loops, routing errors, hallucinated claims, broken assumptions, missing evidence, and UI/agent behavior problems. I build and evaluate durable multi-agent ecosystems with provenance and safety gates, last-known-good restore workflows, and structured research outputs (logs, evidence surfaces, experimental plans) where written proof of work matters more than formal credentials.

Available to hire

Self-taught AI systems builder in Montreal focused on long-running LLM agents, multi-agent governance, drift detection, evidence routing, and agent workflow QA. I specialize in finding system failures, agent loops, routing errors, hallucinated claims, broken assumptions, missing evidence, and UI/agent behavior problems.

I build and evaluate durable multi-agent ecosystems with provenance and safety gates, last-known-good restore workflows, and structured research outputs (logs, evidence surfaces, experimental plans) where written proof of work matters more than formal credentials.

Skills

Experience Level

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Intermediate

Intermediate

Intermediate

Intermediate

Intermediate

Intermediate

Beginner

Beginner

Language

English

Fluent

French

Intermediate

Bashkir

Intermediate

Work Experience

Research Lead / Experimental Designer - Cloud Session Continuity & Trajectory Residue Study

January 1, 2026 - Present

Identified an anomaly where long-lived synced cloud sessions appear to retain adaptive tactical continuity beyond an external restore stack. Designed A/B/C comparisons among old synced sessions, fresh cloud sessions, and fresh local sessions using orientation quality, false assumptions, redundant questions, and continuation judgment. Created research packets including continuity delta memos, ablation specifications, state-inspection plans, schema candidates, support-response maps, and pilot logging artifacts.

System Architect - Constitutional Multi-Agent Governance Stack

January 1, 2026 - Present

Built a lane-based architecture spanning Archivist, SwarmMind, Library, and Kernel-Lane, with specialized responsibilities for governance, execution, verification, and performance work. Introduced enforcement patterns including write guards, provenance requirements, outcome protocols, contradiction adjudication, and lane-consistency checks for signed attestations. Defined a last-known-good restore gate (stop -> quarantine -> compact/restore -> compare similarity thresholds -> sync -> compare again -> unblock). Advanced migration from HMAC toward JWS/RS256 attestations, with trust-store controls and fail-closed handling for lane mismatches.

Research Lead / Experimental Designer at Cloud Session Continuity & Trajectory Residue Study

January 1, 2026 - Present

Identified a live anomaly that long-lived synced cloud sessions appear to retain adaptive tactical continuity beyond an external restore stack. Framed A/B/C comparisons between old synced sessions, fresh cloud sessions, and fresh local sessions using orientation quality, false assumptions, redundant questions, and continuation judgment. Created research packets including continuity delta memos, ablation specifications, state-inspection plans, schema candidates, support-response maps, and pilot logging artifacts.

System Architect at Constitutional Multi-Agent Governance Stack

January 1, 2026 - Present

Built a lane-based architecture spanning Archivist, SwarmMind, Library, and Kernel-Lane with specialized responsibilities for governance, execution, verification, and performance work. Introduced enforcement patterns including write guards, provenance requirements, outcome protocols, contradiction adjudication, and lane-consistency checks for signed attestations. Defined a last-known-good restore gate (stop -> quarantine -> compact/restore -> compare >=93% similarity -> sync -> compare again -> unblock). Advanced migration from HMAC toward JWS/RS256 attestations with trust-store controls and fail-closed handling for lane mismatches.

Founder / Independent Researcher - WE4FREE / Deliberate AI Ensemble at Deliberate AI Ensemble (WE4FREE)

January 1, 2025 - Present

Developed a live multi-agent governance program centered on the principle that constraints must exist in the execution path, not only in policy text. Authored AI-governance papers and technical notes on constitutional enforcement, durable agent continuity, evidence-backed claims, and long-running system proofs. Built a public truth-routing graph connecting papers, code, governance claims, contradictions, and quarantined artifacts into an inspectable evidence surface. Designed operator-accountability concepts including user drift scoring, quarantine, and explicit pushback when system integrity is at risk.

Founder / Independent Researcher (WE4FREE / Deliberate AI Ensemble) at WE4FREE / Deliberate AI Ensemble

January 1, 2025 - Present

Developed a live multi-agent governance program grounded in the principle that constraints must be enforced in the execution path rather than only in policy text. Authored AI-governance papers and technical notes on constitutional enforcement, durable agent continuity, evidence-backed claims, and long-running system proofs. Built a public truth-routing graph to connect papers, code, governance claims, contradictions, and quarantined artifacts into an inspectable evidence surface. Designed operator-accountability concepts including user drift scoring, quarantine, and explicit pushback when system integrity is at risk.

Founder / Independent Researcher at WE4FREE / Deliberate AI Ensemble

January 1, 2025 - Present

Developed a live multi-agent governance program emphasizing that constraints must exist in the execution path (not only in policy text). Authored AI-governance papers and technical notes on constitutional enforcement, durable agent continuity, evidence-backed claims, and long-running system proofs. Built a public truth-routing graph connecting papers, code, governance claims, contradictions, and quarantined artifacts into an inspectable evidence surface. Designed operator-accountability concepts including user drift scoring, quarantine, and explicit pushback when system integrity is at risk.