AI Solution Architect · Supply Chain & SAP · Houston, Texas

I design AI that fits the way the business actually runs.

Twenty years inside procurement, SAP ECC and supply-chain operations at billion-dollar companies across three countries. Today I lead Analytics & Market Intelligence for the Supply Chain of a leading North American energy infrastructure company. In parallel, I have architected and built 20+ AI solutions, and the private AI lab they run on.

  • 20+years in Supply Chain, Analytics & SAP
  • 3countries
  • $Bbillion-dollar companies: TC Energy, Vale, AmBev, Gerdau, ThyssenKrupp
  • 20+AI solutions architected and built in just over a year

Where I built my career

Billion-dollar companies across

  • TC Energy2016 – now
  • thyssenkrupp (ThyssenKrupp CSA)2007 – 2015
  • Vale2007
  • AmBev2006 – 2007
  • Gerdau2002 – 2005

Why this combination

Most AI projects fail between the model and the business process. I have lived on both sides.

Data scientists rarely know how a purchase order really flows. Business leaders rarely build agents. I have spent two decades on the business side of that gap and the last year building the AI side, which is exactly the seat an AI Solution Architect has to sit in.

The process

Over two decades of experience, I know how supply chains really run

Sourcing, contracts, procure-to-pay, inventory and logistics: from the buyer's desk to leading the analytics function at a leading North American energy infrastructure company. I know where the exceptions, workarounds and controls live, which is exactly where AI projects succeed or stall.

The systems & data

SAP since 2002: the person leaders call for supply-chain data

SAP MM certified and called "a true SAP guru" by colleagues. Pulled into divestitures, spin-offs and enterprise transformation programs because I know where the data comes from and what it really means. AI is only as good as the ERP data underneath it.

The AI

20+ AI solutions architected and built in just over a year

A complete stack: governed gateway, agent runtime, deterministic memory, retrieval, optimization and an evidence engine, running on a six-machine AI lab I built myself. Designed to be governed, observable and honest about its limits. Built, not drawn on slides.

Energy · Steel · Mining · Beverages

AI is only as good as the ERP data underneath it. I have spent twenty years making that data trustworthy.

Selected work

Business transformation, grounded in data

Clean data, clear processes and trusted leadership: the foundation every enterprise AI program depends on, and the part most AI teams underestimate.

Spend analytics · Enterprise

Turning enterprise spend data into a decision-ready asset

Delivered a view of enterprise spend the organization had never reported before, on a challenging deadline with half the team out. Recognized with a Gold award.

  • Enterprise data
  • Data limitations
  • Executive reporting

Process & automation · SAP ECC

Redesigning and automating procure-to-pay in SAP

Cleaned, measured and automated a core ERP process at national scale, giving leadership a process they could measure and trust: exactly the groundwork AI agents need.

  • Procure-to-pay
  • SAP ECC
  • Power Automate

Leadership · Strategy

Building a market-intelligence capability

Selected through a competitive process to lead a team of business analysts that gives Supply Chain leadership a forward view of supply trends, risks and opportunities.

  • Team leadership
  • Supply risk
  • Decision support

All work →

AI systems

B1Brain: an AI ecosystem I architected end to end

About two dozen repositories, designed like an enterprise AI platform: governance at the edge, agents in the middle, memory and knowledge underneath, all connected through open standards (MCP, OpenAI-compatible APIs, A2A). It is the kind of platform I would design for your organization, and I have already built it.

  • Gateway & governanceConnect: one front door for every model, with spend receipts
  • Agent orchestrationClaw agent OS · Framework protocols · Cortex · Code
  • MemoryAgentMemory (deterministic, compliance-grade) + Memory (LLM-assisted)
  • Optimization & knowledgeTurboAccuracy · DaVinci evidence engine · Wiki: 33,000+ repositories, 1B+ lines of code analyzed
  • Industry verticalsAEC engineering calculations · Foresight research · WebsiteFactory

Flagship · Evidence

DaVinci

A deterministic engine that turns any codebase into an evidence pack AI agents can trust, with no LLM guessing. It powers the Wiki: 33,000+ open-source repositories and 1B+ lines of code analyzed.

Rust · 14 crates · Active development

Flagship · Memory

AgentMemory

Deterministic, compliance-grade memory for AI agents, with no LLM in the loop. One binary that plugs into the major coding agents (Claude, Codex, Gemini and more) over MCP.

Rust · 16 crates · Pre-release

Flagship · Optimization

TurboAccuracy

An optimization engine (QUBO/Ising) that runs on consumer GPUs instead of specialized hardware. Used to decide what context an AI model sees: a supply-chain-style selection problem, applied to AI.

Python · JAX/CUDA · Active development

Flagship · Gateway

Connect

One OpenAI-compatible front door that routes work across local models, cloud APIs and agent tools, across six execution surfaces, with spend receipts so there is no silent spend.

Python + Rust · Pre-release

The full ecosystem: Claw, Framework, Cortex, Wiki, AEC and more →

The lab

Most architects rent their AI. I built my own lab to run it.

A private, six-machine compute cluster on an encrypted mesh network. It runs local models, agent fleets and evaluations around the clock, with no data leaving my control: the same discipline an enterprise needs for data privacy, cost control and vendor independence.

Server cabinets with amber and teal cabling in a dark room
  • 6machines on one private network
  • ~250 GBof GPU-addressable memory for models
  • ~600 GBof system memory
  • ~85 TBof storage for data, models and evidence

Flagship GPU workstation

Intel Core Ultra 9 285K · NVIDIA RTX 5090 (32 GB) · 256 GB RAM · 14 TB

Training, fine-tuning and the heaviest inference.

2× AI mini-supercomputers

AMD Ryzen AI Max+ 395 · 96 GB unified memory each

Large local models at low power, side by side.

Data & services server

Dell Precision 5820 · Xeon W · RTX 4060 Ti (16 GB) · ~68 TB storage array

Datasets, model library, evidence archive and services.

Development laptop

AMD Ryzen 7 · RTX 4070 Laptop · 64 GB RAM

Where I design, code and orchestrate the fleet.

Low-power test node

NVIDIA GTX 1050 Ti (4 GB)

Kept on purpose to test how models and agents perform on modest, entry-level hardware, not just on a flagship GPU.

One private network

Encrypted mesh across every machine, wherever it sits

Workloads move to the right hardware without exposing anything to the internet.

How I work, in colleagues' words

The person leaders call when the data has to be right and the deadline is fixed.

  • 30peer recognitions
  • 27colleagues
  • 2Gold awards
  • 3Silver awards
  • 9years, 2018–2026
“He quickly got up to speed on the context and data limitations and developed a new view that had never previously been reported by Supply Chain… Bernardo re-prioritized, put in the hours, and delivered.”
Gold award, 2026 · first-ever enterprise spend view
  • Ownership

    “You demonstrated true leadership and took accountability to deliver critical visibility for leaders across all BUs; including rearranging vacation time to make sure this task was completed.”
    Silver award, 2021
  • BI & analytics

    “Bernardo was optimistic and resolute to resolve the problem, and we could not have done it without him!”
    Silver award, 2025
  • SAP & data

    “Greatly appreciate you taking the time troubleshoot the data mining within SAP and teaching me afterwards! You're a true SAP guru.”
    Bronze award, 2021
  • Enterprise programs

    “Your great work, flexibility, and responsiveness ensured that Supply Chain deliverables remained off the critical path to close.”
    Bronze award, 2020 · corporate divestiture
  • Enterprise programs

    “Your deep understanding of our systems, willingness to collaborate and jump to action were key in supporting multiple workstreams…”
    Bronze award, 2023 · enterprise focus project
  • Enterprise programs

    “I know your knowledge is valuable in the company and you are spread thin, but you are always there to help when needed.”
    Bronze award, 2023 · corporate spin-off
  • Analytics

    “Your guidance and advice with our reports and data has been invaluable… Your skill and work sense has certainly made our work life better!!”
    Peer award, 2024 · Transformation Management Office
  • Governance

    “Thank you for bringing the dashboard and your privacy concerns to our attention, great catch!”
    Colleague, 2026
  • Assurance

    “Bernardo – you are a wizard. Thank you so much for this, this report will be extremely helpful in our assurance activities.”
    Colleague, 2024
  • Analytics

    “Bernardo's analytical skills and Supply Chain knowledge greatly increased the quality of the outputs from the working group.”
    Peer award, 2026
  • Ownership

    “You handled everything calmly and professionally and delivered a final product that met all of the defined attributes.”
    Peer award, 2024 · sourcing event data
  • SAP & data

    “Your SAP system knowledge and willingness to share ideas has also created efficiencies related to system reporting and order tracking.”
    Peer award, 2018
  • SAP & data

    “You are always very helpful, a great team player and provide input and suggestions based on your knowledge in SC and of the data.”
    Peer award, 2022
  • BI & analytics

    “Your responsiveness and commitment ensured we had the insights we needed.”
    Peer award, 2025 · KPI dashboard
  • Leadership

    “Amazing! Thanks for your continued leadership Bernardo!”
    Leadership, 2025
  • Leadership

    “We value your partnership and One Team approach!”
    Peer award, 2026
  • Leadership

    “Let's go!!!!! Appreciate the one team approach and all the effort”
    Leadership, 2026
  • Ownership

    “You are always very responsive and it does not go unnoticed. You are also very detail oriented and your work is done perfectly every time.”
    Bronze award, 2018
  • Procure-to-pay

    “Thank you for all your efforts in reducing our invoice exceptions by over 200.”
    Bronze award, 2019
  • SAP & data

    “Being a consistent sounding board and technical expert for eCommerce transactions.”
    Peer award, 2020
  • Procure-to-pay

    “Going above and beyond to assist with eCommerce PO Cutover”
    Peer award, 2019
  • Procure-to-pay

    “I really appreciate your attention to detail, communications on PO status and willingness to pull reports together – often at short notice…!”
    Peer award, 2019
  • Ownership

    “Thank you for your sense of urgency, your attention to detail and your collaboration… Love having you in the supply chain family and your contribution is noticed!”
    Peer award, 2022
  • SAP & data

    “The information you have provided us with has been extremely helpful and will allow for us to get one step closer to a solution”
    Peer award, 2018
  • Procure-to-pay

    “Bernardo's timely issuance of POs makes my job as electric unit fleet manager much more efficient.”
    Peer award, 2018
  • Enterprise programs

    “Thank you for your contributions in making the Sales and Use Tax Project successful!”
    Peer award, 2020
  • Ownership

    “I know it was a lot of work and a tight timeline. Thanks!”
    Peer award, 2021
  • Colleagues

    “you are one of the smartest at TC”
    Colleague, 2026

Excerpts are verbatim from peer recognitions and messages; names withheld.

Career

From the buyer's desk to architecting AI

Each step built on the last. Scroll to build the stack: twenty years of foundation, with AI on top.

  1. 012002–2010 Brazil

    Procurement

    The buyer's desk

    Buying, contracts and negotiation at Gerdau, AmBev, Vale and ThyssenKrupp CSA. SAP key user from day one: I learned where the exceptions, workarounds and controls live.

    GerdauAmBevValethyssenkrupp

  2. 022010–2020 BrazilCanada

    SAP & BI

    The systems and data behind buying

    BI and systems coordination at ThyssenKrupp CSA, SAP MM certified in 2011, then Senior Buyer and Procure-to-Pay analyst at TC Energy in Calgary: master data, KPIs and automation at national scale.

    thyssenkruppSAPTC Energy

  3. 032020–2025 CanadaUSA

    Analytics leadership

    Analytics for the whole supply chain

    Lead Supply Chain Business Analyst: digital solutions, cost-benefit and TCO, and the person leaders called for SAP and supply-chain data through a divestiture, a spin-off and enterprise transformation programs.

    TC Energy

  4. 042025–now USA

    AI architecture

    Leading it, and building the AI layer

    Selected through a competitive process to lead Analytics & Market Intelligence in Houston, while architecting 20+ AI solutions and the private six-machine lab they run on.

    TC Energy

Read the full experience →

Hiring an AI Solution Architect?

Don't take my word for it. Paste the role description and get an honest read on whether I fit: evidence, gaps and a recommendation. That includes "not a fit" when it's true.

See if I fit your role →
Portrait of Bernardo Mendonça

Contact

Let's talk about your AI roadmap

I'm looking for an AI Solution Architect role where business process, enterprise data and AI meet, and where the goal is AI that actually changes how the business runs.

Email
acmbernie@gmail.com
LinkedIn
linkedin.com/in/bm2022
GitHub
github.com/bernie-b1stAI
Location
Houston, Texas, United States