AI-Powered Job Market Intelligence Platform
Automatically collect job postings from LinkedIn, Indeed, and Glassdoor — then let a local large language model extract the intelligence hidden inside: skills in demand, seniority patterns, salary ranges, remote trends, and company hiring signals. All delivered as a live analytics dashboard.
LinkedIn · Indeed · Glassdoor — collected automatically on schedule via Apify
Llama 3.1 8B via Ollama — extracts skills, seniority, salary, work setting, sponsorship
Skills trends, salary ranges, seniority distribution, company signals — updated every run
Deploy on your terms — local server, private cloud, or on-premise. Your data never leaves your environment.
MarketLens connects to LinkedIn, Indeed, Glassdoor, and other major job boards through a reliable harvesting layer. Searches are fully configurable — keywords, location, portals, max results, and lookback window. The pipeline runs on a schedule and deduplicates postings automatically, so your database grows cleanly over time.
Every collected job posting passes through a locally-hosted large language model (Llama 3.1 8B via Ollama). The LLM reads the full description and extracts structured fields: required skills, seniority level, salary range, work setting, employment type, role category, development type, and sponsorship requirements. No cloud API needed — the model runs on your hardware.
An always-live dashboard summarises the entire market at a glance: top skills bar chart, role category pie, seniority distribution, salary ranges, timeline of postings per day, top hiring companies, most active locations, remote vs hybrid vs on-site breakdown, and employment type split. Click any chart element to jump to the filtered jobs list instantly.
Browse every collected posting in a sortable, filterable table with 15+ search parameters: keywords, seniority, role category, skill, source, location, date range, salary floor/ceiling, work setting, employment type, and custom flags. Save your filter sets as defaults. Flag individual jobs as starred, applied, rejected, or archived. Right-click any row to act.
The pipeline pulls job postings from configured portals on schedule. Each run collects up to 500 fresh postings, skipping exact duplicates already in the database.
Each new posting is passed to the local LLM. The model reads the full description and returns structured JSON: skills, seniority, salary, work setting, role category, and more.
Enriched data is aggregated into the analytics layer. Dashboard charts, stats cards, and trend lines update automatically — no manual refresh required.
Search, filter, flag, and drill into the data. Click a chart to filter the jobs table. Open original postings directly. Export via SQL for deeper analysis.
At the core of MarketLens is a locally-hosted Llama 3.1 8B model running via Ollama. Unlike cloud-based AI services, the model operates entirely within your infrastructure — your job market data never leaves your environment.
The LLM does what would take a human analyst hours per day: it reads the natural language of each posting and produces structured, queryable fields. No rules engines. No regex patterns. No hand-coded taxonomy. The model understands context — it knows that "5+ years with PyTorch" means Senior, that "competitive compensation" needs to be flagged as undisclosed, and that "AWS, Python, and Spark" map to distinct skill categories.
Discuss a DeploymentThe LLM identifies every technical and soft skill mentioned in the posting — normalised to a consistent vocabulary so charts aggregate meaningfully across thousands of jobs.
Junior · Mid · Senior · Lead · Principal · Executive — classified from context, not just title keywords. The model reads years of experience, scope of responsibility, and reporting structure.
Min and max salary extracted from any format the posting uses — hourly, annual, range, or approximate — and normalised to an annual figure for consistent benchmarking.
Remote · Hybrid · On-site classified from job description context. Visa sponsorship availability extracted and flagged — critical signal for workforce planning and candidate fit.
User registration with email verification and role-based access control. Admin and standard user roles. Public registration can be disabled after initial team onboarding.
Optional TOTP-based 2FA compatible with Google Authenticator, Authy, Microsoft Authenticator, and any standard TOTP app. Per-user control from the Settings panel.
Deploy on a local server, private cloud (AWS, Azure, GCP), or on-premise infrastructure — all with a single command. Scale workers and API instances independently. Nginx load-balances traffic automatically, supporting multi-server deployments for large teams.
Talk to us about deploying MarketLens for your team or organisation.