Problem Solver + Designer + Team Player = Engineer
Pathfinder:
A career sensemaking tool for a world where paths are no longer linear.
Pathfinder helps people reflect on their professional direction by turning messy, emotional career thoughts into structured insights, gentle hypotheses, and small, testable experiments, without telling them what they “should” do.
The Problem
As technology accelerates and new roles emerge faster than education systems can adapt, more people feel lost in their careers, even highly capable, motivated ones.
​
Traditional career tools:
-
assume linear paths
-
over-index on titles and resumes
-
optimize for “fit” instead of meaning
-
give prescriptive advice without context​
​
In reality, career decisions are deeply human: shaped by energy, values, environments, uncertainty, and lived experience. People don’t need answers, they need sensemaking.
What I Built
Github Repo: https://github.com/AntonellaMasini/pathfinder-api
Pathfinder is an AI-powered reflection tool designed to support career sensemaking without overreach.
Users write freely about their experiences: what energizes them, what drains them, what they want more (or less) of.
​
Pathfinder then:
-
Extracts structured signals (energizers, drainers, values, skills, work styles)
-
Reflects those patterns back in natural language using tentative, non-prescriptive framing
-
Proposes “path hypotheses” not job recommendations, but archetypal directions to reason about
-
Suggests concrete micro-experiments: small, real-world actions that can be tried within a week
-
Adds voice playback so reflections can be heard, not just read, making the experience more human and accessible
To reduce harm and overconfidence, the system includes an internal evaluation layer that checks for:
-
overreach
-
prescriptive language
-
vague or unrealistic actions
and retries generation only when necessary.​
Why this Approach
Built with:
Pathfinder is intentionally not a career recommender.
​
Instead, it’s built around a few core beliefs:
-
People are experts on their own lives.
-
AI should help humans think more clearly, not decide for them.
-
Reflection feels different when spoken aloud.
-
Small experiments are more honest than big predictions.
​
This project sits at the intersection of:
-
applied AI systems
-
human-centered design
-
ambiguity, identity, and decision-making
-
FastAPI + Python
-
OpenAI models for structured reflection and synthesis
-
ElevenLabs for text-to-speech
-
React frontend, prototyped with Lovable and refined through custom logic and integration
Next Steps:
Pathfinder is an MVP , but it opens space for deeper questions:
-
How can AI support self-understanding without authority?
-
What does responsible career guidance look like in an AI-driven economy?
-
How does voice change how people process feedback?
-
Can sensemaking tools be adaptive, ethical, and emotionally aware?
​​
Future directions could include memory, longitudinal reflection, voice input instead of text, or research into how people act on hypotheses over time, but the core goal remains the same: help people think, not tell them what to do.