Alan Gaudet
Internal P&G team member (6+ years; promoted to Team Second in 2025) building demonstration agentic-AI systems on my own time. IWS Autonomous Maintenance and Daily Management Systems on the plant floor, paired with three end-to-end AI projects over the last year.
Team Second, Manufacturing Operations
Procter & Gamble · Belleville, ON
Promoted from Manufacturing Technician (2019–2025) to Team Second (2025).
- Equipment owner under IWS Autonomous Maintenance: CIL (Clean-Inspect-Lubricate) routes, defect tagging, and condition-based monitoring across pneumatic and hydraulic systems. Currently acting team lead during team leader's leave.
- Drive AM standards on owned equipment, nearly eliminating targeted stops through root-cause analysis and centreline management.
- Run Daily Management System (DMS) routines on shift: tier-1 board reviews, shift handoffs, loss tracking, and upward escalation to maintenance and engineering.
- Zero safety incidents on shift in the last year. Author SOPs adopted on shift; train new team members on AM standards and equipment-specific procedures.
- Prior technician tenure built deep operational knowledge of P&G IWS systems and Allen-Bradley PLC control stacks — the foundation under the industrial-AI work (see Rung Reader).
Independent AI & Systems Development
Self-directed
See selected projects below for technical detail. TDD with ruff + mypy strict gates on every merge.
Team Manager
HGS Canada Inc. · Belleville, ON
Led a team of 20 agents for 6+ years.
- Consistently top-quartile for CSAT, first-contact resolution, and QA.
- Team Manager of the Month for performance multiple times during tenure.
- Built Excel tracking tools adopted by the broader team.
Earlier
Rogers Communications · Convergys
Rogers Non-Live Consultant, Toronto (2008–2012): top 10% on all targets, led department-wide process-improvement initiatives. Convergys Associate Trainer, Belleville (2006–2008): delivered training for 15–25-person new-hire cohorts.
supply-chain-agents
Three specialist agents coordinating through a typed LangGraph state machine with human-in-the-loop approval.
- Unit + real-LLM integration tests gating every merge in CI; REST API with SSE streaming.
- Self-hosted Langfuse v4 Docker Compose stack (web, worker, Postgres, ClickHouse, Redis, MinIO).
- Bicep IaC patterns modelling Container Apps + ACR deployment.
- Two-tier output guardrails: regex hard-block + LLM judge with hard timeout, fail-open on every error path.
rung-reader
AI-powered Allen-Bradley PLC ladder logic troubleshooter. Pure-function L5X walker; LLM only narrates.
- Unit + integration test suite gating every merge. On-premise deployment, zero cloud dependencies.
- Graceful degradation without LLM, PLC, or Redis. Pure-function L5X walker; LLM only narrates.
- Mock-swappable pycomm3 driver for full CI without real hardware.
- arq + Redis async worker for proactive anomaly detection on live tag streams.
irontide
From-scratch Rust BitTorrent engine targeting libtorrent-rasterbar parity. Async actor architecture with backpressure.
- 31 BEP protocol extensions: v1 + v2, Merkle verification, hybrid torrents, DHT, PEX, uTP, UDP tracker.
- 17-crate workspace; unit + integration + property tests; clippy-clean and lint-strict in CI.
- Async actor architecture on tokio select! loops with backpressure.
- Pluggable crypto (AWS-LC / ring / OpenSSL) and disk I/O (POSIX / io_uring / IOCP / mmap). Throughput benchmarked vs. libtorrent.
Secondary Diploma
Nicholson Catholic College, Belleville, ON
Ongoing self-directed technical education (2024–Present) in agentic AI, Rust systems, and cloud-native design.