The setting
is a transportation platform that connects riders with rideshare, bikes, scooters and other ways to get around. gives the drivers who provide those rides tools for earnings, payments, safety and support.
Safety outreach needs incident context that can also identify the reporter.
A driver's opportunity to answer a report conflicts with the reporter's confidentiality when a small detail identifies the ride.
Who you’d be doing this for
“I've driven twenty trips since Saturday. Just saying someone felt unsafe doesn't tell me what I'm supposed to explain.”
Mamadou Baffoe · Rideshare driver · a character in this brief
Receives a request to explain a reported incident before a human associate reviews the case.
What is at stake
Safety messages for 2,400 drivers may identify the people who reported them. You must weigh usable incident context against disclosures that cannot be recalled once today's messages leave the queue.
Why it isn’t already fixed
Every obvious fix costs something else. That’s the part you’d have to decide.
- reporter confidentiality vs actionable incident context
- outreach coverage vs case-level scrutiny
- consistent evaluation vs ambiguous identification
- release qualification vs immediate human calls
Why Lyft
LangGraph safety case management turns confidential reports into direct driver contact within a service handling millions of human and AI interactions.
Written with these in mind
Not your kind of problem? 76 more at Lyft, 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 Lyft. They are listed as sources, not as openings. The lines under a posting are its own words, where it meets this brief.
- Machine Learning Engineer, Safety and Customer Care AI (opens in a new tab)
- Design and build AI-powered support agents and end-to-end agents for safety case management using LangGraph or equivalent agentic frameworks.
- We fine-tune and align open-source models, build AI-powered support agents, and develop end-to-end AI agents for safety case management, systems that reason over complex, high-stakes cases and drive them to resolution.
- Data Scientist - Algorithms, Mapping (opens in a new tab)
- Senior Software Engineer, Forecasting & Real-Time Optimization (opens in a new tab)
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 people starting something and never quite finishing it.
Some agent summaries omit a customer’s recorded opt-out request.
You must weigh durable context protection against retrieval cost and release risk.
Reviewed answers sometimes cite the right document while giving contradictory instructions.
You must weigh stronger retrieval coverage against a release path people can trust.
Three of 240 reviewed cases merge facts from separate providers despite an 11-point completeness gain.
You have to weigh a one-way routing decision against the cost of keeping two expensive paths alive.
The same setting, from another role
Each is set around Lyft’s Lyft for Drivers, written for a different role. They’re separate briefs, not the same scenario.
Airport wait-time guidance is staying stale after airport communication changes, sending drivers into queues that no longer match the app.
You must decide what the system can rank automatically and what engineers must verify before acting.
Drivers are opening completed trips and still contacting support to ask what changed their pay.
You have to weigh a useful explanation against the stability and confidentiality of earnings configurations.
Safety notifications are flagging ordinary trips after an app release changed incoming sensor signals.
You must weigh immediate protection for drivers against losing coverage for genuinely unsafe driving.
Other problems with a similar purpose
In each of these, as in this one, the aim is to build machine-learning systems that earn trust by staying explainable where their errors carry real consequences.
An audit found sensitive fields in 7% of shared artifacts that manual review missed, and the classifiers disagree on the edge cases.
You have to build evidence for a production commitment you cannot easily reverse.
One threshold sends very different numbers of drivers into coaching in each region, and reviewers there agree less often.
You have to balance fair treatment of drivers against enough coverage to improve coaching.
Confirmed takeover messages are reaching guests, while legitimate pre-check-in threads are also being blocked.
You must weigh fraud prevention against a guest losing the instructions needed to enter a booked stay.
