Report 02 of 09All work
Career guidance, English and Hindi / Built, waiting on launch blockers

Bodhmarg

Career guidance for Indian students: a conversation, five assessments and a roadmap a parent can read too.

Next.js 16React 19SupabaseClerkStripeUpstashZodPlaywrightOpenRouter
Fig. 1 / Bodhmarg in use
109
assessment questions, up from 6
5
assessments, fully bilingual
~400
career and guide pages generated
100
Lighthouse score on career pages, all four categories

01Summary

Bodhmarg takes a student through an intake chat and five assessments, then writes a personalised roadmap: three careers, each with milestones at one, three and five years. A parent can be linked and sees their own view. Everything works in English and Hindi.

Around it sits a library of career profiles and guides, built for search, so that a student who has never heard of the product can still land on something useful.

02Why I built it

Most career advice a student in India receives is a relative's opinion. The tools that exist are either a single quiz or a counsellor who costs more than the family wants to spend. I wanted something in between that a parent would also trust.

03How it works

  1. Step 1

    Consent

    Role and age are asked before anything else.

  2. Step 2

    Intake

    A streaming chat that remembers the session and gathers context.

  3. Step 3

    Assess

    Interest, aptitude, personality, work values and emotional intelligence. Aptitude uses real problems with verified answers, not self-rating.

  4. Step 4

    Interpret

    Insights written in both languages from the scored results.

  5. Step 5

    Roadmap

    Three careers with one, three and five year milestones.

  6. Step 6

    Share

    A parent view, linked to the student.

Fig. 2The question bank, by assessment

Four to five questions per dimension, interleaved, with reverse-keyed personality items scored correctly. The server requires at least 80 percent answered.

04Decisions

DecisionReason
Free models do the legworkA router sends routine work to free models. The expensive model receives a compact brief, not the whole conversation.
Gender-neutral HindiThe question bank was rewritten so that no student is addressed with the wrong form.
Plain names for traitsThe results say emotional sensitivity where the textbook says neuroticism. A sixteen-year-old reads this.
A feature checklist in the repositoryEvery feature and fix is recorded in the same commit that makes it, with a date.

05What broke

06What is next

A production sign-in instance on a custom domain, model credits for the Hindi pages, and a redesign of parent linking.

Hire meBook a call