Twenty years of professional software experience across gift-card processing, fraud ML, payroll integrations, and document architecture — plus a hands-on side practice in firmware analysis and embedded Linux. Working with LLMs since ChatGPT launched in 2022, and now helping organizations move to agentic processes safely. Available for system-modernization, cloud-migration, agentic-transformation, and full-stack engagements in 2026.
Take one service on its own or combine several. Most engagements start with a short written assessment, then move to fixed-scope delivery or an ongoing retainer.
Move how your organization works toward agentic processes one layer at a time, with controls in place before anything gets autonomy.
Hand-deployed servers become containers on managed Kubernetes, without a big-bang cutover.
Proof · ENG-A · ENG-C
A DigitalOcean or AWS cluster your team can reason about, from ingress to the egress IP your vendors allowlist.
Proof · ENG-A · OpenSensor platform
Batch files and scheduled tasks on someone’s server become versioned, logged, re-runnable pipelines.
Proof · ENG-A · ENG-C
SIS, HR/payroll, LMS, CRM, and billing systems that agree with each other — over APIs instead of file drops.
Proof · ENG-A · ENG-B
Accounts that are created, grouped, and retired by rule rather than by ticket.
Proof · ENG-A
The response steps you would do by hand at 2 a.m., written down and turned into a one-click pipeline.
Proof · ENG-A · ENG-C · CVE work
Logs, traces, and reporting in one place, so “is it broken?” has a fast answer.
Proof · ENG-A · ENG-B
LLMs used where they save real time, with validation and a human reviewing what they produce.
Proof · ENG-A · Agentic program · OpenSensor platform
An inherited codebase with no documentation becomes one you can change with confidence.
Proof · ENG-A · ENG-D
Checkout and money movement built by someone who has sat on PCI and SOC 2 audit teams.
Proof · ENG-C · ENG-D · OpenSensor store
An outside look at the boot chain and firmware of the hardware you ship or deploy.
Proof · OSS-03 · CVE work
After the migration, someone who knows the system keeps it patched, monitored, and moving forward.
Proof · ENG-A
An organization is built like an onion. The outer layers are easy to change, but many of the inner ones depend on the old world: file drops, manual handoffs, and systems only one person understands. Agents cannot safely work on top of those dependencies, so the work is to peel them back in short iterations. Each step is controlled and measured, and each one can be undone.
Drafting, research, summarizing. Assistants help people; nothing acts on its own.
Tickets, docs, code review. Agents propose changes and people approve them.
Scoped, read-first connections (MCP) to the systems of record, with every call logged.
Defined pipelines where agents do the steps and gates decide what ships.
File drops, manual handoffs, undocumented systems. Rebuilt with APIs, docs, and tests so agents can work on them safely.
Outer layer = lowest risk · core = most old-world dependencies
Find what each process depends on, and which of those dependencies belong to the old world.
Run one process with agents in dry-run or propose-only mode next to the existing one.
Add least-privilege credentials, approval gates, audit logs, and a rollback path.
Give more autonomy only where the results support it, then move to the next layer.
↻ Repeat, one layer deeper
One engagement touched almost every service listed above. Here is where the college started, and what runs there today.
Founder & Principal Engineer
Cumberland Center, Maine
Software engineer with twenty years of professional experience architecting production systems for fintech, e-commerce, higher education, and enterprise clients. Founded OpenSensor Engineering LLC in mid-2025 after 3.5 years at a fintech company, leading payroll integrations and document architecture.
Earlier career: a decade in the digital gift-card industry at CashStar and Blackhawk Network — Python/Django and Java services that powered millions of retail transactions, an introduction of Terraform and ECS across roughly 50 services, and three consecutive years on PCI and SOC 2 audit teams. Prior to that, an SDE I role at Amazon writing tooling for hardware load balancers, and a BSCE from the University of Illinois (cum laude, 2008) with a minor in International Engineering — Italian Studies.
Side practice in firmware and embedded Linux — about three years deep, but the output is real: a maintained Python Packaging Authority project, an open-source NN runtime built from a clean reverse-engineering, and coordinated CVE disclosures on Ingenic SoC boot ROMs (BSides Maine 2026 talk).
Service decomposition, message-driven workflows, REST/GraphQL APIs, and the React/Next.js layer on top. Delivered for fintech and e-commerce platforms processing real volume.
Python · Django · Java · React · Next.js · PostgreSQL · Redis
On-prem-to-cloud migrations on AWS and DigitalOcean Kubernetes. Terraform-driven infrastructure introduced across roughly 50 services in a previous role; ECS, EKS, and DO clusters in production.
AWS · DigitalOcean · Kubernetes · Terraform · ECS · Docker
Legacy monolith decomposition, version uplift, and replacement of file-based integrations with API-first contracts. Strategic, incremental, and verifiable in production. Workflow-engine-driven orchestration when the steps need to be auditable.
Microservices · REST · GraphQL · Conductor (Orkes) · Celery
Gift-card processing, transaction-level fraud ML (logistic regression, xgboost, Pandas-driven feature work), and three consecutive years on PCI / SOC 2 audit teams. Incident manager (escalated L3 on-call) for three years.
PCI · SOC 2 · Pandas · xgboost · Logistic Regression · L3 On-Call
Pragmatic ML in production: model deployment, automated transaction categorization, document classification, and edge inference (200× speedup on an embedded NN accelerator). Pandas-heavy data pipelines feed everything upstream.
Pandas · xgboost · PyTorch · ONNX · Data Science · LLMs · Edge AI
Elasticsearch / Kibana monitoring stacks, ingress hardening, and proactive threat detection. Document architecture built on S3 + Elasticsearch in a previous role.
Elasticsearch · Kibana · Terraform · Ingress
Side practice — about three years deep but the output is real: firmware analysis, secure-boot review, ISP / sensor bring-up, and reverse-engineering work that has produced CVEs and an open-source NN runtime.
C · Rust · Buildroot · OpenWRT · GDB · IDA
Tier-2 New England institution (abstracted)
Architected the on-prem-to-cloud migration of an institutional platform onto DigitalOcean Kubernetes. Navigated seven major version uplifts in sequence, replaced legacy file-based integrations with API-first contracts, and automated the jobs around them. See the case file above.
Kubernetes · Java · Jenkins · REST · Active Directory · LLMs
Fintech
Led the Gusto payroll integration enabling S-corp salary processing in-app. Subsequently built the in-app tax-organizer workflow on Orkes Conductor, used across three consecutive tax seasons. Quarterly tax-estimate calculator and automated transaction categorization.
Python · Conductor · S3 · Elasticsearch · Stripe
Blackhawk Network
Senior Staff. Introduced Terraform and converted ~50 services to ECS on Docker with auto-scaling. Built fraud ML (logistic regression / xgboost) at the transaction layer. Incident manager (escalated L3) for three years; PCI + SOC 2 audit team three years running.
Terraform · ECS · Python · xgboost · PCI · SOC 2
CashStar
Backend re-architecture eliminating Tomcat in favor of Python APIs validating requests via RabbitMQ. Python 2→3 conversions, Django uplifts, Celery introduction, Redis as memcached replacement (with a locking/throttling library used widely in-house).
Python · Django · Celery · Redis · RabbitMQ · Java
OpenSensor sister property
Lightweight network-video-recorder for low-resource hardware. Now also a managed cloud product.
Python Packaging Authority
Long-term maintainer. Release engineering, dependency-resolver work, and triage of an OSS project relied on by tens of thousands of Python developers.
thingino-accel
Open-source NN runtime built from a clean reverse-engineering of an embedded NN accelerator. MXUv3 SIMD acceleration and ORAM memory optimization. 200× faster inference than the vendor stack.
Open firmware for IP cameras
Upstream contributions to Thingino, the open-source firmware project for affordable IP cameras built on Ingenic SoCs.
A short note describing the problem — the system, the constraint, the deadline — gets you a fast read and a real reply. If a 30-minute call makes sense after that, we'll set one up.
