The setting
's combines devices and software to monitor safety in physical operations. Its dash cams, gateways, and sensors run AI that detects distracted driving and helps prevent collisions as they happen.
A passing retry does not settle whether a filter failure needs a fix.
Repeated investigation consumes feature capacity, while unstable test data weakens the evidence used to route failures.
Who you’d be doing this for
“Yesterday I changed a filter and got a different set of events. My supervisor asked which list we should coach from.”
Petra Todorovic · Fleet Safety Manager · a character in this brief
Uses filtered event review to prepare supervisor follow-ups and is participating in the upcoming customer evaluation.
What is at stake
Engineers keep rebuilding failed filter tests before they can judge a regression. You decide whether AI can propose routing after a passing retry, or whether that result must first go back to an engineer.
Why it isn’t already fixed
Every obvious fix costs something else. That’s the part you’d have to decide.
- repeatable fixtures vs realistic shared state
- broad failure coverage vs bounded replay capacity
- removing test interference vs diagnosing genuine regressions
- concise triage vs visible uncertainty
Why Samsara
Safety event review filters determine which incidents safety managers see when choosing where to follow up.
Written with these in mind
Not your kind of problem? 43 more at Samsara, or browse every organization.
Where the organization context comes fromThe postings this brief was written from, and the reporting read before it.
The organization context comes from public information. The situation and characters are written for the brief.
Job postings it was written from
Public descriptions of the work at Samsara. They are listed as sources, not as openings. The lines under a posting are its own words, where it meets this brief.
- Senior Software Engineer II (opens in a new tab)
- You don't need to train models, but you know how to take AI capabilities and turn them into products customers love.
- You'll participate in customer interviews, collaborate with product managers and customer success, and use that context to build products that solve real problems operators face every day.
- Sr. Firmware Engineer II (opens in a new tab)
- Every model we ship runs in the real world, on real roads, protecting real drivers.
- We build AI-first, using AI tools to develop, test, and ship features to customers quickly.
- Sr. Software Engineer I (opens in a new tab)
- The Safety Triage & Insights team builds data-driven safety experiences that surface actionable insights, engage drivers directly, and help fleets focus resources where they matter most.
- The team is deeply invested in AI, LLMs, and emerging agentic technologies, both in the products we build and in how we engineer them, enabling rapid iteration and highly personalized, insight-driven user experiences.
- Firmware Engineer (opens in a new tab)
- Securing embedded systems end to end - secure boot, firmware signing, cryptography, and key management that survives contact with manufacturing.
- Sit in reviews with our electrical engineers and help choose the right SoC, amount of flash/RAM, and combination of sensors, then prove those choices out on prototypes that reach real customers in months, not years.
This is the setup. The work is inside.
Running it puts you in the room: the full situation and its constraints, stakeholders who push back in their own words, and the decisions that are yours to make. What you produce becomes a Day One Plan — work you can show someone instead of describing.
Where similar work appears
Other live briefs connected to this one, and what connects them.
Similar problems at other organizations
Like this one, each of these is about the everyday tools people rely on silently breaking.
Regression triage takes 45 minutes before engineers can decide whether a failed test blocks an application release.
You have to decide where the drafted failure brief stops and the engineer's own call on cause and release begins.
Cloud incidents are taking longer to diagnose because build, route, and edge evidence must be reconstructed by hand.
You must decide what AI can rank safely and what an engineer must verify before a fix moves forward.
Workflow failures take 45 minutes to turn into an engineer-approved investigation summary.
You have to decide where the AI stops reading CloudWatch and Datadog evidence and drafting failure paths, and where an engineer takes over the root-cause call and any change to jobs or retries.
The same setting, from another role
Each is set around Samsara’s Worker Safety, written for a different role. They’re separate briefs, not the same scenario.
Safety managers marked 210 of 600 reviewed events as connected to another event.
You must weigh a simpler encounter review against the chance of merging situations that require separate follow-up.
Tomorrow, 1,200 vehicles leave with collision warnings that cannot reliably change for 14 days.
You must weigh additional hazards caught against nuisance warnings using evidence that cannot all be independently checked before departure.
A dash camera pilot produces 36 extra warnings on clips that pass model evaluation.
You must weigh a corrected model export against reverting firmware and giving up its latency gain.