---
title: "Senior Multimodal Document Intelligence Engineer"
slug: "senior-multimodal-document-intelligence-engineer"
id: "c68e6cc6-e81b-438e-8ba6-4ea3fb7d45fd"
department: "Applied AI"
department_code: "APP"
location: "New York, NY"
workplace_type: "Hybrid"
employment_type: "Full-time"
compensation: "$200k – $285k (annual package)"
is_remote: false
is_featured: false
headcount: 1
canonical_url: "https://jobs.talent247.co/jobs/senior-multimodal-document-intelligence-engineer"
markdown_url: "https://jobs.talent247.co/jobs/senior-multimodal-document-intelligence-engineer.md"
apply_endpoint: "https://jobs.talent247.co/api/public/jobs/senior-multimodal-document-intelligence-engineer/apply"
schema_endpoint: "https://jobs.talent247.co/api/public/jobs/senior-multimodal-document-intelligence-engineer/form-schema"
webmcp_tool: "apply_for_role"
updated_at: "2026-11-15T17:35:10.022Z"
content_signal: "search=yes, ai-input=yes, ai-train=no"
---

# Senior Multimodal Document Intelligence Engineer

## Position Telemetry

- **Reference:** `SENIOR-MULTIMODAL-`
- **Practice / Sector:** Applied AI (`APP`)
- **Location:** New York, NY
- **Workplace Type:** Hybrid
- **Employment Type:** Full-time
- **Compensation:** $200k – $285k (annual package)
- **Canonical HTML URL:** https://jobs.talent247.co/jobs/senior-multimodal-document-intelligence-engineer

## Role Dossier & Specification

THE CLIENT & MANDATE
Our client is a flagship enterprise AI company serving elite global law firms, private equity sponsors, and corporate legal teams from New York and San Francisco. Every downstream reasoning workflow in their platform depends on lossless ingestion of tens of millions of complex real-world documents—ranging from heavily redacted scanned PDFs and multi-page nested cap-table spreadsheets to handwritten margin annotations and cross-referenced deal exhibits.

They have engaged Talent247 to recruit a Senior Multimodal Document Intelligence Engineer in New York, NY. Sitting at the intersection of vision-language modeling and high-throughput ingestion infrastructure, the successful candidate will build layout-aware VLM parsing pipelines and hierarchical document graphs that preserve every footnote, clause boundary, and tabular cell with near-perfect fidelity.

KEY RESPONSIBILITIES
• Train, fine-tune, and deploy compact Vision-Language Models (VLMs) and layout-detection transformers specialized for complex multi-column agreements, dense financial tables, redlines/track-changes, and degraded scans.
• Architect hierarchical structural chunking and cross-reference resolution graphs that link defined terms, schedules, signature pages, and embedded exhibits across multi-thousand-page deal rooms.
• Engineer GPU-accelerated ingestion pipelines (vLLM / TensorRT-LLM + async Rust/Python workers) capable of parsing tens of thousands of uploaded pages within seconds of data-room ingestion.
• Build multimodal bounding-box grounding pipelines so every generated legal or financial claim highlights the exact pixel coordinates and table cell in the source document viewer.
• Construct adversarial document-parsing benchmarks covering rotated scans, multi-page spanning tables, watermarks, multilingual appendices, and complex mathematical formulas.
• Optimize cost-quality routing between ultra-fast classical OCR/PDF-tree heuristics and heavy VLM visual-token passes on ambiguous pages.

TECHNICAL ENVIRONMENT & OFFERING
Stack: Python, Rust, PyTorch, Vision-Language Models (Qwen-VL/Pixtral-class custom fine-tunes), vLLM, PDFium/Poppler, Apache Arrow/Postgres, Kubernetes. Workplace: Hybrid in New York, NY. Compensation: $200,000 – $285,000 base salary plus significant equity ownership and full US benefits.

## Core Requirements

ESSENTIAL QUALIFICATIONS
• 4+ years of ML engineering experience in multimodal vision-language models, document AI, OCR/layout understanding, or visual information extraction.
• Proven experience fine-tuning and serving VLMs, multimodal transformers, or layout-aware vision encoders in PyTorch and vLLM/TensorRT.
• Deep familiarity with PDF internals, coordinate bounding-box alignment, complex table-structure recognition (HTML/Markdown/JSON serialization), and reading-order reconstruction.
• Strong systems engineering skills to scale batch and real-time GPU document pipelines with tight p95 latency and memory budgets.
• Disciplined evaluation mindset building cell-level TEDS, edit-distance, and downstream retrieval-recall benchmarks.

PREFERRED / BONUS EXPERIENCE
• Experience distilling frontier multimodal models into sub-7B vision-language parsers running at hundreds of pages per second per GPU.
• Background processing financial filings (10-K/S-1s), M&A virtual data rooms, clinical trial PDFs, or patent specifications.
• Proficiency in Rust or C++ for high-speed PDF rasterization and structural pre-parsing.

## Application Schema & Direct Apply (`apply_for_role`)

Candidates and authorized AI agents on behalf of a candidate can submit an application via the declarative WebMCP form tool `apply_for_role` on https://jobs.talent247.co/jobs/senior-multimodal-document-intelligence-engineer or via HTTP `POST https://jobs.talent247.co/api/public/jobs/senior-multimodal-document-intelligence-engineer/apply` (`application/json` or `multipart/form-data`). Non-browser HTTP agents must include header `X-Talent247-Intake-Token` obtained from `GET https://jobs.talent247.co/api/public/intake-token` (or automatically returned in `GET https://jobs.talent247.co/api/public/jobs/senior-multimodal-document-intelligence-engineer/form-schema`).

### Form Fields

- `full_name` (text, required): Full name
- `email` (email, required): Email
- `phone` (phone, optional): Phone
- `location` (text, optional): Location
- `linkedin_url` (url, optional): LinkedIn URL
- `cover_letter` (textarea, optional): Cover letter
- `consent` (checkbox, required): I consent to my data being processed for this application
- `cv_file` / `cv_base64` + `cv_filename`: Upload a CV in .pdf or .docx format (<= 10 MB) or provide an `https://` LinkedIn profile URL in `linkedin_url`.

### Alternative: Talent Pool Intake (`join_talent_pool`)

If this mandate is not an exact match, register once in the Talent247 Talent Pool at https://jobs.talent247.co/talent-pool (`https://jobs.talent247.co/talent-pool.md`) via WebMCP tool `join_talent_pool` or `POST https://jobs.talent247.co/api/public/talent-pool`.
