At 11pm in a tier-2 city, a first-generation student has the degree. What they do not have is a mentor who will still pick up.
That is the local problem I am building for.
I am Abhishek Kumar. I am in Google Cloud Gen AI Academy APAC Cohort 3. This is my Meet the Builders story. Early-stage is allowed. A real local problem is not optional.
Meet the Builders is a campaign to find, document, and celebrate 1,000 builders across Asia Pacific who are using Google AI tools to solve real, local problems.
The solution could be a crop-disease alert for smallholder farmers in Indonesia. An AI-assisted triage system for an underfunded clinic in Bangladesh. A multilingual chatbot that helps migrant workers in Malaysia navigate legal rights.
These stories deserve to be told. Mine is about career access in India.
The degree without the map
In India, career advice still concentrates in a few metro campuses and a few LinkedIn circles. First-generation students in places like Indore, Ranchi, Coimbatore, Guwahati, and Nashik often finish college with a certificate — and no one who can tell them what “good” looks like in year one.
The placement cell closes. The WhatsApp groups go quiet. Family wants a stable job, not a slogan about “breaking into product.” Generic chatbots fill the silence with invented salaries, fake company processes, and advice that does not survive a Hindi–English mixed interview.
That is not a content shortage. It is an access problem.
They get stuck on the same three questions:
What am I actually hiring for — PM, BA, operations, data — and what does “good” look like in year one?
How do I turn college projects into a portfolio a hiring manager will trust?
What should I practice this week — not “learn product management” as a slogan?
I have watched campus talent get filtered out for lack of translation, not lack of ability. The work is there. The map is not.
What I am building
PathSetu — path plus setu, a bridge — is a student-facing AI mentor on Cloud Run. It answers in plain language, grounded in a curated career knowledge base: role definitions, portfolio rubrics, interview drills, weekly plans. Not the open internet.
If the question needs a human — a specific college placement cell, mental health, a scholarship office — it says so and stops.
The point is not to replace a good teacher. It is to give a first-generation student a repeatable weekly plan at 11pm, in the language they actually use, without waiting for a metro mentor to reply.
Architecture (production path)
Student → ADK + Gemini → RAG knowledge pack → Cloud Run → grounded answer, or human handoff.
Ask
Student writes in plain language — Hindi, English, or both.
Reason
ADK + Gemini run on Cloud Run — the production path.
Ground
Retrieve from the curated pack. Never the open internet.
Decide
Cite and answer, or stop and hand off to a human.
Cite the note. Answer.
Role maps, portfolio rubrics, weekly drills.
Stop. Hand off.
Placement cell, counsellor, scholarship office.
ADK + RAG
Gemini + BigQuery / MCP
Cloud Run productivity agent
Mapped to the Cohort 3 labs:
| Cohort 3 lab | PathSetu equivalent |
|---|---|
| Personalized customer experience (ADK + RAG) | Mentor agent that retrieves role maps, portfolio checklists, and interview drills from approved notes |
| Location intelligence (Gemini + BigQuery / MCP) | Where to run campus workshops first: city and college clusters with high demand and weak mentor supply |
| Operations productivity agent on Cloud Run | Placement-cell / campus-club assistant for session plans, follow-ups, and weekly practice queues |
Rules I will not break
This is where most career chatbots fail. They sound confident. They are not grounded.
- No invented salaries, headcount, or “guaranteed placement” claims
- Cite the retrieved note when the answer is process-specific
- Escalate when the user needs a human counsellor or an official college office
- Keep student PII out of prompts; treat chat logs as sensitive
If PathSetu cannot ground the answer, it does not guess. Guessing is how first-generation students get hurt.
Why this matters in APAC
Mentor access is uneven across this region the same way it is uneven inside India. The Meet the Builders brief is explicit: celebrate the work that never makes a keynote — farmers, clinics, migrant workers, campus corridors.
A grounded agent on Cloud Run will not close every gap. It can sit beside a student at 11pm and refuse to hallucinate a future they cannot use.
That is the story I am submitting.
Watch the story
1080p walkthrough, about 4 minutes. Same story as this write-up. Drive / YouTube link goes here after upload (anyone with the link).
If you are a builder in APAC using Google AI on a local problem, join Meet the Builders. Early-stage is allowed. The problem has to be real.