Nebula
JobPilot (Nebula) targets enterprise HR talent acquisition teams and tech job seekers. What goes wrong today: Legacy Applicant Tracking Systems (ATS) rely on exact-string keyword matching, systematically misclassifying qualified candidates with unstructured CV formats or non-standard skill phrasing. Recruiters spend 15+ hours weekly manually scanning resumes and conducting preliminary screens, while job seekers receive zero actionable feedback on skill gaps.
Preliminary score
- Value for the user14.5 / 25
A specific problem and a clear outcome from solving it.
- Prototype and use of AI15.5 / 30
A working core scenario, and what AI actually contributes to the result.
- Quality testing8.0 / 20
Tests, examples of failures, and comparison with the current approach.
- Feasibility5.5 / 15
Data requirements, running costs and a clear next step.
- Originality3.0 / 10
A meaningful difference from common solutions.
What the judges wrote
- GPT-6 AstraPreliminary51/100
- Strengths
- Your CV-to-profile-to-job-matching flow addresses a recognizable problem: translating varied resume language into relevant skills and opportunities. Your architecture gives AI a substantive extraction and matching role rather than a decorative one. Your testing write-up identifies concrete integration failures, including oversized API payloads and hydration errors, and explains how you addressed them.
- To improve
- Your enterprise case needs a clearly demonstrated recruiter workflow, especially since your deck describes employers as a future audience. Include a worked CV/job example with extracted skills, matching rationale and interview output to substantiate the end-to-end scenario. Report labeled results from your 50+ CV tests, including extraction failures and matching accuracy against a keyword baseline, and compare timing for equivalent tasks. Replace the template README with project-specific documentation covering data sources, CV privacy and retention, cost per applicant and a scoped pilot.
- Claude Opus 5.5Preliminary42/100
- Strengths
- Your write-up names a real pain point: keyword-based ATS filters reject qualified candidates who phrase skills differently, and job seekers get no feedback on their gaps. You have a deployed live demo with a clear flow (PDF upload, skill extraction, job matching, AI mock interview), so the core scenario appears to run on an LLM rather than being slides only. You were honest about engineering failures, such as Groq 413 payload errors and rate limits, and you described concrete fixes like truncation and strict JSON schemas.
- To improve
- Your audience shifts between documents: the submission targets enterprise HR teams, while the deck targets students and graduates and calls employers a 'future audience'. For this enterprise track, pick one buyer and one workflow. Your testing covers build and API bugs, not matching quality. Report results from the 50+ CVs, for example agreement with recruiter shortlists, cases where the score was wrong, and a measured keyword-ATS baseline, instead of an unmeasured '<3 seconds vs 10 minutes' claim. Your README is still the default create-next-app text. Add setup steps and an architecture note, and in the deck cover cost per screened CV, data sources and a concrete pilot step.
The team
Aytac Rahimli
designer
Gurbanzadeh Hormuz BahruzLead
developer
Şəbnəm Orucova
developer