Joe Poznanski

Summary

Principal Software Engineer with 15+ years shipping mission-critical aerospace and high-fidelity simulation software (Otto Aerospace, Blue Origin). I build agentic AI systems and developer automation that turn ambiguous operational goals into reliable, test-backed software: tool-using agents, eval/guardrail pipelines, and self-healing workflows that accelerate requirements-to-code while improving safety and correctness. Deep hands-on delivery across Rust / C / C++ / Python, distributed systems, and operator-grade dashboards.

Agentic AI & automation highlights

• Built tool-using agent workflows that automate triage to plan to implementation loops with human-in-the-loop review. The result is repeatable, test-backed changes instead of one-off prompt outputs.
• Created AI-assisted QA pipelines (test scaffolds, evidence packaging, regression summaries) that shorten feedback cycles on safety- and mission-critical systems.
• Developed ops copilots for log/telemetry summarization and anomaly correlation across distributed services, cutting time-to-diagnosis during integration and drills.
• Standardized prompt/playbook libraries and deterministic guardrails so teams can adopt AI workflows safely at scale.

Selected systems

Experience

1. Otto Aerospace

   Principal Software Engineer - SuperNatural Vision Program

   Aug 2025 - Present

   Ft. Worth, TX

   • Architected and shipped an 8K video-wall system: synchronized playback across a 4-node Mac Mini cluster (~50 GB/s aggregate) with sub-frame latency, driving immersive executive demos.
   • Built a Rust/tokio orchestration backend (async SSH dispatch, JSON-RPC playback control) with <100 ms command round-trip; and a Node.js/React operator dashboard with <16 ms UI updates.
   • Operationalized agentic AI workflows (Claude/Codex + tool-use) for triage, code-gen, and refactoring across Rust/Python/Node. The resulting changes are reviewable, test-backed, and wired into CI/CD.

2. Blue Origin

   Senior Simulation / Aerospace Operations Software Engineer

   May 2022 - Aug 2025

   Cape Canaveral, FL

   • Built a Rust-based Simulated Avionics Layer delivering 500k+ telemetry signals at greater than 10 Hz from a single 32-core server. Maintained control-loop jitter below 1 ms at 100 Hz.
   • Automated test & compliance: Python/pytest + GitLab CI running 800+ unit/functional/HIL tests nightly, auto-generating DO-178C / AS9100 evidence; regression cycles 3 days → 2 hours.
   • Shipped the first-gen Operations Training Simulator (MVP in 18 months); trainee setup time −70%, NPS 93/100.

3. Avenger Flight Group

   Director of Simulation & Software Engineering

   Jun 2014 - May 2022

   Dallas, TX / Las Vegas, NV

   • Scaled the company 11 → 200+ staff across three countries; hired and mentored the first 30 simulation/software engineers.
   • Designed a real-time C++17 / VxWorks / ZeroMQ data-bus (<120 ms) now fielded on 18+ aircraft simulators; earned 100% first-pass FAA approvals via CI/CD + automated DO-178C artifacts.
   • Led technical delivery of a $45M multi-year simulator refresh (A320, 737 MAX, ATR), lifting narrow-body market share 1% → 13%.

4. SIMCOM Training

   Flight Simulation Supervisor & Project Manager

   Feb 2011 - Jun 2014

   Orlando, FL

   • Built an enterprise C++/SQL analytics platform ingesting Qualification Test Guide data from 12 flight-training devices, auto-generating compliance logs.

Technical skills

Languages

Rust · C · C++ (17/20) · Python · Java · TypeScript/JS · SQL · Bash · PowerShell

AI / ML

Agentic systems & multi-agent orchestration · tool-using agents · evals/guardrails · Claude · OpenAI/Codex · PyTorch · TensorFlow

Systems & Tooling

tokio · Boost · Docker · GitLab/Jenkins CI/CD · AWS · Terraform · Ansible · Linux/RT · Grafana/Loki · React · Node.js

Aerospace & Standards

DO-178B/C · AS9100 · ARP4754A/4761 · FAA Part 60 · MBSE · HIL simulation · DAL-A/E