Data Science Resume Feedback and Semantic Job Recommendation System

This is a submission for the Redis AI Challenge: Real-Time AI Innovators.

What I Built

Career Code Advisor is a Redis‑powered, AI‑augmented job search app. It matches candidates to roles using hybrid search (full‑text + vectors), provides résumé‑aware recommendations, and surfaces real‑time engagement (live favorites + a rolling 60‑minute trend).

Highlights

• Paste or upload your résumé → instant matches + short coaching feedback.
• Faceted filters (location, company, skills) that feel instant.
• Autocomplete & Ctrl+K quick launcher for speedy navigation.
• Semantic fallback when text search returns few/no results.
• Save jobs (⭐) and see live favorites in the upper right as well as real time increases in decreases in a grafana dashboard.

Demo

• Live app: https://careercodeadvisor.com (The UI exposes Ctrl+K, Saved Jobs, “Live favorites” and “60min trend”, plus a Résumé Matcher.) :contentReference[oaicite:6]{index=6}
• Source: https://github.com/ericmariasis/ai-career-advisor

How I Used Redis 8

• RedisJSON as primary store for job docs:
  • Example shape: { title, company, location, skills[], text, embedding }
• RediSearch for full‑text + faceted filters:
  • Title/company as TEXT, skills and location as TAG, salary as NUMERIC.

• Vector KNN for semantic search & “more‑like‑this”:
  • Embedding dim: [fill: e.g., 1536], metric: COSINE, algorithm: [HNSW or FLAT].
• Semantic cache for résumé feedback/skill extraction:
  • Cache key: embedding of input; reuse if similarity ≥ [fill threshold, e.g., 0.90].
• Real‑time UI surfaced via live favorites and a 60‑minute trend on Dashboard.

⚡ Try It Now (60 seconds)

1. Open https://careercodeadvisor.com.
2. Type a natural query like: python and toggle a filter (e.g., Location).
3. Try a quirky query (typo or rare phrase) to see semantic fallback suggestions.

1. Click Résumé Matcher, paste 6–8 lines from your résumé, and hit Match & Advise.

1. ⭐ Save a job, then open Dashboard to see Live favorites and the 60‑minute trend update. A green line will go up as you save favorites and go down as you remove favorites.

🛣️ What’s Next

• Profile‑aware ranking (learn from saves/clicks).
• Alerts (new roles matching saved skills).
• Multi‑embedding backends; offline batch re‑indexer.
• Deeper analytics: Redis Streams → dashboard panels.

Why Redis?

With Redis 8 as the real‑time data layer, one stack powers search, vectors, caching, and live activity — keeping responses snappy and the UX visibly real‑time.

Solo project by **Eric Mariasis* (@ericmariasis).*