Build AI‑Powered Outage‑Focused Log Analysis Agent
Budget: $30 – $250 USD
We’re seeking an experienced freelancer to create a lightweight, end‑to‑end AI agent that automatically ingests heterogeneous log files, pinpoints outages, and summarizes root‑cause insights. The agent must pull logs from Azure, normalize timestamps across varying formats/timezones, accept natural‑language queries (e.g. “Check for outage on Jun 10 around 10 AM”), detect common failure patterns (OutOfMemory, DB errors, etc.), leverage charts/metric images to focus analysis around spikes, and then produce a concise summary of findings. Finally, it should update an existing JIRA ticket and optionally open new tickets when follow‑up is needed.
Key Responsibilities
Log Ingestion & Normalization
Fetch and extract logs from Azure Blob/File storage.
Parse diverse timestamp formats and timezones per file.
Natural‑Language Interface
Map free‑text prompts to UTC time windows and filtering rules.
Outage & Anomaly Detection
Scan for predefined error signatures (OUOM, deadlocks, service crashes, etc.).
Correlate across multiple log types (runtime, audit, DB, data‑layer, report).
Chart/Spike Correlation
Consume associated chart images or metrics, detect anomalies, and narrow log scan windows.
AI‑Powered Summaries
Integrate OpenAI (with a flag to swap in alternative LLMs) to generate clear, actionable incident summaries.
JIRA Integration
Programmatically post results to a given JIRA ticket and create new tickets if criteria are met.
Scalability & Robustness
Efficiently handle both small and very large log sets (streaming, batching).
Graceful error handling and clear logging/documentation.
Deliverables
Codebase – well‑structured, tested Python modules.
Documentation – setup guide, usage examples, configuration options.
Demo – end‑to‑end walkthrough using sample logs & queries.
Timeline
Duration: ~1–2 weeks
Key Responsibilities
Log Ingestion & Normalization
Fetch and extract logs from Azure Blob/File storage.
Parse diverse timestamp formats and timezones per file.
Natural‑Language Interface
Map free‑text prompts to UTC time windows and filtering rules.
Outage & Anomaly Detection
Scan for predefined error signatures (OUOM, deadlocks, service crashes, etc.).
Correlate across multiple log types (runtime, audit, DB, data‑layer, report).
Chart/Spike Correlation
Consume associated chart images or metrics, detect anomalies, and narrow log scan windows.
AI‑Powered Summaries
Integrate OpenAI (with a flag to swap in alternative LLMs) to generate clear, actionable incident summaries.
JIRA Integration
Programmatically post results to a given JIRA ticket and create new tickets if criteria are met.
Scalability & Robustness
Efficiently handle both small and very large log sets (streaming, batching).
Graceful error handling and clear logging/documentation.
Deliverables
Codebase – well‑structured, tested Python modules.
Documentation – setup guide, usage examples, configuration options.
Demo – end‑to‑end walkthrough using sample logs & queries.
Timeline
Duration: ~1–2 weeks