Skill Stack

Inspiration

Our inspiration stems from our firsthand experience with a flawed recruiting system. Applicants often send out countless internship applications with little to no success, while recruiters are overwhelmed by the sheer volume of resumes. Furthermore, these resumes frequently contain unverified claims, making it incredibly difficult for companies to efficiently identify true talent.

What it does

SkillStack is a decentralized "Proof of Performance" platform that replaces static resumes with verifiable, AI-validated credentials. It works by immersing users in live, interactive video roleplays, such as high-pressure sales pitches or HR conflict resolutions, where a reactive AI avatar evaluates their logic, tone, and de-escalation skills in real-time.

How we built it

We deliver realistic, soft-skill-focused interview simulations using LiveAvatar. During the interaction, we capture key data points and process them using the Google Gemini API to generate objective assessment scores for individual skills. After securely storing the records in MongoDB Atlas, we leverage the Solana blockchain to mint a non-fungible token (NFT) of the final assessment. This ensures that every user’s verified score is immutable and universally verifiable.

Challenges we ran into

Integrating a high-stakes live simulation presented significant technical hurdles, primarily in balancing real-time responsiveness with blockchain finality. We faced substantial latency issues while synchronizing the live AI video avatar with multimodal analysis, as any delay in the feedback loop broke the immersion of the roleplay. On the backend, we battled race conditions where concurrent database writes from the live evaluation engine competed with the asynchronous Solana transaction signing process, occasionally leading to state mismatches. Finally, managing Solana’s integration required careful handling of RPC congestion and transaction retry logic to ensure that every performance hash was reliably anchored to the ledger without forcing the user to wait for on-chain confirmation before completing their session.

Accomplishments that we're proud of

We are incredibly proud of successfully bridging the gap between high-fidelity AI interaction and data integrity. Our biggest milestone was achieving a low-latency, multimodal evaluation engine that can analyze a user's performance in real-time without interrupting the flow of a live video roleplay. We also engineered a seamless Solana integration that abstracts away the complexity of the blockchain, allowing users to mint a "Proof of Skill" credential with a single click. Seeing a subjective human interaction -- like a complex negotiation or a difficult HR conversation -- translated into a structured, cryptographically verifiable data point on-chain is a major step toward our vision of a trustless, global labor market. Beyond the technical build, we are proud to have created a platform that gives individuals true ownership over their professional reputation, turning "hidden" soft skills into a portable and liquid asset.

What we learned

Building SkillStack taught us that the hardest part of "Agentic Education" isn't the AI itself, but the orchestration of state and synchronization. We learned how to manage the delicate "handshake" between real-time inference and blockchain finality, specifically how to handle asynchronous Solana transactions without degrading the user's live video experience. Developing this project deepened our understanding of multimodal data integrity, as we had to ensure that the AI’s qualitative analysis remained consistent and tamper-proof from the moment of capture to its archival on the ledger. Ultimately, this journey shifted our perspective on professional identity: we realized that the future of hiring lies in portable, cryptographic proof of action rather than centralized, static credentials.

What's next for Skill Stack

Our immediate roadmap focuses on expanding SkillStack from a single-player simulation into a multi-sided ecosystem. We plan to develop "Custom Skill Blueprints," allowing organizations and universities to upload their own unique rubrics and proprietary case studies to create branded certification tracks. To enhance the simulation’s fidelity, we aim to integrate low-latency emotional intelligence (EQ) tracking that can detect subtle micro-expressions and vocal tremors, providing even deeper insights into a candidate’s stress management. Finally, we intend to launch a Recruiter Verification Portal on Solana, enabling employers to search for verified talent based on specific performance hashes, effectively turning the Solana ledger into the world’s most transparent and high-signal job market.