Independent project

ClearSOC

An AI-assisted security operations workspace for investigating application security events and deciding what to look at next.

Independent project · Under active design and development

Security operationsClickHouseAI-assisted investigation
Analyst workflow

Move from a spike to a decision.

  1. 01 / Detect

    Spot a spike

    See that security activity has changed enough to warrant a closer look.

  2. 02 / Scope

    Identify affected applications

    Understand where the increase is happening before diving into individual events.

  3. 03 / Investigate

    Inspect the evidence

    Explore event data with queries or ask follow-up questions in natural language.

  4. 04 / Review

    Consider next steps

    Review suggested actions as decision support; an analyst remains responsible for what happens next.

Example investigation questions

  • “Which applications saw the biggest increase in blocked requests?”
  • “What changed during the spike?”
Architecture

A telemetry-to-investigation path

  1. 01

    Ingest

    Akamai DataStream telemetry is collected from the configured object-storage feed. The source implementation loads compressed event files for processing.

  2. 02

    Store and query

    Processed events are stored in ClickHouse for aggregation and investigation.

  3. 03

    Enrich and investigate

    Additional context supports triage. Analysts can explore the data with SQL and use a natural-language assistant to ask follow-up questions. Specific enrichment fields and sources are not listed here.

  4. 04

    Review analysis and briefings

    Security-analysis reporting is present in the implementation. The briefing concept is to summarize activity worth attention; this page does not claim a fixed report format or generation schedule.

My contribution

Designed and built independently.

I’m designing and building ClearSOC as an independent project. Its codebase includes a ClickHouse ingestion and query path, analyst investigation workflows, a natural-language assistant interface, and security-analysis reporting. Feature availability is still in development.

AI-generated recommendations are inputs for analyst review. A confidence indicator is a review cue, not a measured probability of correctness. ClearSOC does not automatically change security controls through this recommendation flow.

Project conversation

Interested in the problem space?

Get in touch to discuss the project and its direction.

Contact Alexandre