When a wrong answer costs someone their career.
HumanLens is partnering with the Weitzman Institute at Moses/Weitzman Health System to build and govern an AI Career Pathway Engine for frontline healthcare workers. The project is funded by the GitLab Foundation AI for Economic Opportunity Fund.
The problem, on two levels
Frontline healthcare workers do not lack talent. They lack visibility. Millions of people in support roles already hold valuable, transferable skills but cannot see a clear way up. The questions they carry are concrete:
- What jobs am I qualified for?
- What are my transferable skills?
- What additional skills would increase my earning potential?
- Which credentials are actually worth the time and money?
- How long will it take to reach my career goals?
- What pathway is realistic given my family, finances, and work schedule?
Traditional career ladders are static, hard to personalize, and rarely reach the people who need them most.
There is a second problem hiding inside the first. The moment you build an AI system to recommend career moves, you have built a high-stakes model. A recommendation here is not a movie suggestion. It can steer someone to spend two years and real tuition on a path. If the skills data is biased, if the labor-market signal is stale, or if the worker and their coach cannot understand why the model said what it said, the system can quietly send people in the wrong direction and call it guidance.
That is the gap most AI career tools step right past. It is the gap HumanLens exists to close.
Why this is a HumanLens project
HumanLens governs high-consequence AI before it makes decisions about people, and produces the evidence that shows it was done responsibly.
A career engine is a textbook case for that work. It touches livelihoods, it operates in a sensitive population, and it has to be defensible to funders, employers, and the workers themselves. This project lets us do something most governance vendors never get to do: build a real model in a real setting and govern it from the first line of code, rather than reviewing someone else's system after launch.
We are not bolting responsibility on at the end. We are designing for it from the start.
The vision
AI that expands opportunity. We believe artificial intelligence should strengthen economic opportunity, not replace workers.
Our vision is a future where every worker has access to personalized, data-driven career guidance that was once available only through intensive one-on-one coaching. The Career Pathway Engine combines artificial intelligence, labor market intelligence, skills mapping, career coaching principles, and employer workforce data to create personalized pathways toward advancement.
How we work: a design partnership, not a vendor handoff
This is co-development. Each partner closes a gap the others cannot — because responsible AI in a domain this human is not a solo act.
Building the Career Pathway Engine and running it through our responsible-AI methodology, end to end.
Research, market analysis, every stage of the build, and leads pilot testing with frontline workers.
Funds the work through the AI for Economic Opportunity Fund.
Labor-market and skills data.
Cloud infrastructure and generative-AI engineering support.
Five commitments. Five engineering requirements..
HumanLens treats responsible-AI principles as engineering requirements, not values on a poster. For the Career Pathway Engine, each one is a control we check and evidence as the build progresses.
A worker and a coach can see why a pathway was recommended, not just the result.
Recommendations come with the reasoning: skills adjacency, wage signal, time and cost to transition.
The engine informs coaches and workers. It does not replace human judgment, and it is not designed to.
Recommendations rest on labor-market and skills data, validated against how careers actually move in this field.
Labor markets shift, so the system is monitored over time rather than certified once and forgotten.
How the engine works
Step 1: Understand the worker
Current role, experience, skills, education, goals, and the real-life constraints that shape what is possible.
Step 2: Analyze opportunities
Skills adjacency, labor-market demand, wage-growth potential, credential requirements, and time to transition.
Step 3: Generate personalized pathways
Not the next job. The next career.
Step 4: Connect workers to resources
Training, credentials, mentors, employers, and coaching.
Why this is different
Most AI career tools optimize for the next job. The Career Pathway Engine asks a harder question: where can this worker be in two, three, or five years, what sequence gets them there, and which path produces the best return given what they actually have to work with?
Why community health centers
Community health centers are the right place to start. They employ large numbers of frontline workers in exactly the support roles where talent outruns visibility. Their mission is already economic opportunity for the communities they serve, so improving worker mobility is aligned, not added on. And as the research arm of an FQHC network, the Weitzman Institute brings both the population and the domain knowledge to build something that holds up in the real world before it scales anywhere else.
Illustrative: meet Maria
Maria is a patient support associate earning $42,000 at a community health center. She wants to earn more but cannot stop working, because she supports two children.
The Career Pathway Engine maps several routes, estimates the wage gain for each, surfaces local education programs, and recommends the path that best fits her circumstances. Instead of guessing, Maria and her coach work from a personalized roadmap they can both understand.
Maria is an illustrative composite. The project is in pilot, and outcomes are being measured now.
Research and evaluation
The Weitzman Institute leads research and evaluation, with a stated commitment to measuring real economic-mobility outcomes rather than usage metrics. The question is not whether workers like the tool. It is whether they advance, and whether the AI helped them get there. That measurement is underway as part of the pilot.
The evidence base
The approach is grounded in career-mobility and workforce-development research, and in HumanLens CEO Cesar Koirala's doctoral work on leadership and innovation at NYU.
What this work shows
For workforce funders and partners, it shows HumanLens building AI that expands opportunity instead of replacing the people it is meant to serve.
For enterprises evaluating AI governance, it shows something more pointed: HumanLens does not just review models on paper. We are building and governing real, high-consequence AI in a regulated, sensitive domain, and producing the evidence that proves it was done right. The Career Pathway Engine is the methodology in motion, and it is still being built.
Looking ahead
The demonstration starts with healthcare workers. The longer goal is scalable AI tools that help any worker navigate complex career decisions and reach lasting economic mobility, with responsibility built in from the start rather than audited in after the fact.
The same methodology that governs the Career Pathway Engine is what HumanLens brings to high-stakes models inside regulated enterprises..
Whether you're a workforce funder, a healthcare partner, or an enterprise evaluating AI governance — we'd like to talk.