mbilal.works
Available · September 2026
Talk
← Back to Portfolio
Applied AI System & EngineLIVE SYSTEM
Open Source / AI Engineering · Mar 2026 → Present

Multi-Model LLM CV Tailoring Pipeline

Multi-pipeline agentic system that ingests job descriptions, extracts structured experience evidence, scores ATS keyword coverage, and generates tailored DOCX/LATEX CV drafts with real-time SSE progress streaming.

01. Problem & Context

The Operating Challenge

Generic CVs get filtered out by automated Applicant Tracking Systems (ATS) and recruiters. Manual tailoring for every target role takes 1–2 hours per application.

02. Architecture & Solution

Technical Implementation

Engineered 5 pipeline architectures (Employer-First, Validated, Chain, Hybrid, Single) with multi-provider chains. Features exact-match ATS keyword scoring, SequenceMatcher verbatim copy detection, and automated Pydantic repair gates.

03. Measurable Impact

Results & Verification

Evaluated across 700+ model-pipeline-job scenario matrix combinations. Reduces CV tailoring time from 90 minutes to under 30 seconds with 100% document schema validation.

04. Technology Stack
Python 3.14FastAPIsentence-transformersCustom AI AgentsPydanticpython-docx
mbilal.works · personal hub at bilalm.me · Impressum · PrivacyBuilt to a DESIGN.md spec · Bricolage Grotesque + JetBrains Mono · MMXXVI