Candidate Assessment
Forward Deployed Engineer
Demonstrate your ability to analyze a complex manual workflow, design a scalable AI-powered solution, and communicate your technical thinking clearly.
Background
Cadre AI is a growth-stage AI consulting firm that works with mid-market and enterprise companies to design and deploy AI solutions across their operations. Forward Deployed Engineers at Cadre sit at the intersection of client delivery and technical implementation. They are expected to quickly understand a client's existing processes, identify where AI can create leverage, and architect solutions that are practical, integrated, and built to scale.
The ability to break down a messy manual workflow and think clearly about the systems, APIs, and tradeoffs involved in automating it is core to this role.
The Scenario
🏢 Client: Meridian Logistics
Meridian Logistics is a regional freight brokerage headquartered in Dallas, TX. They move approximately 2,000 loads per month across the US, connecting shippers with a network of 800+ carriers. The company has a 40-person operations team managing shipments across phone, email, a TMS, and a patchwork of shared spreadsheets.
Client
Meridian Logistics
Industry
Freight Brokerage
💬 Situation
"We spend half our day on the phone chasing updates. By the time we know there's a problem, the shipper has already called us twice. We have the data somewhere, but nobody can get to it fast enough to do anything useful."
— VP of Operations, Meridian Logistics
After a discovery session with Meridian's ops team, you have documented the following manual workflow they run every day to track active freight shipments and manage exceptions. Review it carefully.
The Manual Workflow: Shipment Tracking & Exception Management
Below are the six steps of Meridian's daily manual workflow. Review all of them carefully — your solution must address the full end-to-end process.
Export and Set Up the Daily Tracking Board
Each morning, an Ops Coordinator manually exports all "In Transit" shipments from the TMS into a spreadsheet. The export includes load ID, origin, destination, scheduled delivery date, assigned carrier, and last status. This snapshot is pasted into a shared Google Sheet — the day's tracking board. It is static from the moment it is exported and is not connected to any live data source.
"Every morning I pull the export, paste it in, and by the time I'm done setting up the sheet, some of those statuses are already an hour out of date."
— Ops Coordinator, Meridian Logistics
Contact Carriers for Status Updates
Three coordinators divide the load list and begin calling or emailing each carrier individually to request a location update and revised ETA. Carriers respond via phone, email, or text with no standard format. Each coordinator manually types responses into the Google Sheet. High-priority loads are flagged with a colored cell highlight, applied by hand. This step takes two to four hours per coordinator depending on carrier responsiveness.
"I have 60 loads today. Some carriers pick up right away, some I'll have to call three times. There's no system — it's just however long it takes."
— Ops Coordinator, Meridian Logistics
Identify Exceptions
Once updates are entered, an Ops Supervisor manually scans the Google Sheet row by row, comparing updated ETAs to scheduled delivery windows to identify exceptions: late deliveries, unresponsive carriers, damaged freight, or loads running more than two hours behind. Exceptions are logged in a separate tab. There is no automated alerting — exception identification depends entirely on how quickly and thoroughly the Supervisor reviews the sheet.
"I'm looking at 200 rows and doing the math in my head. If I miss something or get interrupted, we might not catch a late load until the shipper calls us."
— Ops Supervisor, Meridian Logistics
Notify Shippers of Exceptions
The Supervisor flags exceptions in the sheet and notifies the assigned Customer Success Rep via Slack. The CS Rep then drafts a custom email to the shipper explaining the delay or issue, pulling details from three separate places: the exceptions log, the original load record in the TMS, and carrier notes in the Google Sheet. Email templates exist in Gmail drafts but require manual customization for each situation.
"By the time I've found all the information I need and written the email, 45 minutes have passed. The shipper has probably already called us."
— Customer Success Rep, Meridian Logistics
Escalate and Resolve Unresponsive Carriers
If a carrier is unreachable after two contact attempts, the Supervisor escalates to the Carrier Relations team, who works from a personal contact list maintained in a separate Excel file — not in the TMS. Resolution steps such as rerouting, carrier swaps, or shipper coordination are handled by phone and documented in the exceptions log after the fact. There is no defined SLA for resolution time and performance is not tracked systematically.
"Our best contact info for some carriers isn't in the TMS at all. It's in Mike's spreadsheet on his desktop. If Mike's out, we're guessing."
— Carrier Relations, Meridian Logistics
End-of-Day Reconciliation and Reporting
At the end of day, a coordinator manually updates the TMS with final statuses for each load. Delivered loads are moved to an archive tab in the Google Sheet by copy-paste. Unresolved exceptions are carried forward to the next day's sheet, also by copy-paste. A summary email is sent to leadership with a manually counted breakdown of on-time, delayed, and unresolved exception loads.
"The end-of-day update takes about an hour. Half of that is just moving rows around in spreadsheets. And if I count wrong on the summary, leadership is working off bad numbers."
— Ops Coordinator, Meridian Logistics
Your Task
Meridian wants to explore whether AI can meaningfully reduce the manual burden of this workflow, improve exception response times, and give leadership better visibility into operations. They have not committed to any specific solution or vendor.
Your job is not to build anything. Your job is to think rigorously about what a solution could look like and be honest about where it gets hard.
Analyze the manual workflow above. Design a scalable, AI-powered solution that addresses it. Your analysis must include:
- Map every step of the manual workflow end to end, including all handoffs, tools, and decision points
- Identify which systems, APIs, and platforms a solution would need to integrate with
- Design a scalable solution architecture and explain how each component connects
- Surface at least 3 meaningful limitations or risks in your proposed design
Deliverable: Record a video (up to 10 minutes) walking through your analysis and proposed solution. You may present live, use slides, draw a diagram on screen, or simply talk through your thinking.
Requirements
Video Length
10 minutes maximum
Submission
Share link within 3 business days
You may use slides, a whiteboard, screen share, or simply talk through your thinking on camera. The goal is to demonstrate how you would approach a real client workflow and communicate your technical reasoning clearly.
Submission
Please record a screenshare video and upload it as an unlisted YouTube link, then send the link to:
moriah.davis@gocadre.ai