This is a submission for the Hacktoberfest Weekend Challenge: Build for a Friend
What I Built
Starting out as a software engineer or junior manager is rarely bottlenecked by syntax or code—it's bottlenecked by uncomfortable conversations:
- Pushing back when a senior developer asks for an unrealistic 48-hour sprint turnaround.
- Explaining why tech-debt refactoring is necessary to a product manager focused purely on new features.
- Giving constructive, upward feedback without damaging relationships.
My batchmate and friend was recently dealing with severe anxiety around their upcoming sprint planning meeting. They knew their team's deliverables were overscoped, but they didn't know how to defend the engineering reality without sounding confrontational.
I built PocketMentor for them: a mobile simulator that lets early-career engineers and managers roleplay high-stakes workplace dialogues against an adaptive AI persona and receive actionable feedback on communication tactics.
Demo
Live Demo / Preview: https://github.com/dipanshurdev/pocket-mentor
Walkthrough video:
Code
The source code is open and accessible on GitHub:
PocketMentor – Executive Communication Coach
PocketMentor is a mobile simulator designed for early-career software engineers and junior managers to practice high-stakes workplace conversations (scope pushback, upward feedback, tech-debt defense).
Next Gen Award Track Submission
- Monetization Engine: RevenueCat Test Store SDK
-
Target Entitlement:
pro - Framework: React Native / Expo (TypeScript)
- Persistence: Local-first session history via AsyncStorage
RevenueCat Architecture
- Configuration & SDK initialization:
services/revenuecat.ts - Dynamic Offering & Package resolution:
components/PaywallModal.tsx - Real-time Entitlement checks & feature gating:
App.tsx
Local Setup
npm installnpx expo start -c- Press
wto run web preview or scan QR with Expo Go.
How I Built It
PocketMentor is designed around structured simulation loops powered by open-source AI:
- Scenario Definition: Pre-configured behavioral rubrics (scope pushback, tech-debt defense, upward feedback) establish clear constraints and realistic stakeholder personas.
- Open-Source AI Core: The simulation engine utilizes an open-weight reasoning model (Gemma) to maintain consistent persona fidelity, push back realistically against evasive answers, and evaluate conversational diplomacy.
- Evaluation Loop: At the end of each exchange, the engine critiques tone, clarity, and trade-off articulation rather than generic conversational fluency.
Why Does Open Innovation Matter?
- Psychological Safety & Data Privacy: High-stakes workplace simulations require vulnerability. Users roleplay sensitive scenarios involving internal team dynamics, stressful deadlines, and confidential work context. Relying on open-weight models ensures interactions remain private and out of third-party commercial training pools.
- Custom Behavioral Guardrails: Closed commercial models frequently default to overly accommodating, generic chatbot replies. Open-source models allow fine-tuned system boundaries so the counterparty persona realistically pushes back like an authentic manager or product owner.
- Accessibility: Career development tools shouldn't demand recurring paid subscriptions from students and junior engineers starting their careers.
Friend's Verdict
I handed PocketMentor to my friend to rehearse a real upcoming scenario: defending a sprint delay due to foundational database refactoring.
"Most career advice tells you 'just be diplomatic,' which is useless in the moment. Rehearsing the exact back-and-forth and having the model challenge my excuses helped me restructure my actual talking points for sprint planning."
Prize Categories
- Best Use of Gemma: Configured Gemma as the primary simulation and evaluation engine to drive dynamic stakeholder roleplay and actionable feedback.
Top comments (0)