---
title: "Senior Research Scientist, Flow-Matching Visual Foundation Models"
slug: "senior-research-scientist-flow-matching-visual"
id: "f64d9475-8e67-4c8a-9bab-afa20f205c9c"
department: "AI Research"
department_code: "RES"
location: "Freiburg / Berlin, Germany"
workplace_type: "Hybrid"
employment_type: "Full-time"
compensation: "€125k – €180k (annual package)"
is_remote: false
is_featured: false
headcount: 1
canonical_url: "https://jobs.talent247.co/jobs/senior-research-scientist-flow-matching-visual"
markdown_url: "https://jobs.talent247.co/jobs/senior-research-scientist-flow-matching-visual.md"
apply_endpoint: "https://jobs.talent247.co/api/public/jobs/senior-research-scientist-flow-matching-visual/apply"
schema_endpoint: "https://jobs.talent247.co/api/public/jobs/senior-research-scientist-flow-matching-visual/form-schema"
webmcp_tool: "apply_for_role"
updated_at: "2026-11-15T17:24:52.307Z"
content_signal: "search=yes, ai-input=yes, ai-train=no"
---

# Senior Research Scientist, Flow-Matching Visual Foundation Models

## Position Telemetry

- **Reference:** `SENIOR-RESEARCH-SC`
- **Practice / Sector:** AI Research (`RES`)
- **Location:** Freiburg / Berlin, Germany
- **Workplace Type:** Hybrid
- **Employment Type:** Full-time
- **Compensation:** €125k – €180k (annual package)
- **Canonical HTML URL:** https://jobs.talent247.co/jobs/senior-research-scientist-flow-matching-visual

## Role Dossier & Specification

THE CLIENT & MANDATE
Our client is a globally acclaimed frontier visual intelligence laboratory headquartered in Freiburg, Germany (with a second hub in Berlin), founded by the original researchers behind latent diffusion and high-resolution flow-matching transformers. Backed by tier-one Silicon Valley and European investors at unicorn valuation, they train state-of-the-art foundation models for image synthesis, context-aware visual editing, typography rendering, and generative video.

They have partnered with Talent247 to appoint a Senior Research Scientist, Flow-Matching Visual Foundation Models in Freiburg (or Berlin hybrid). Joining a compact, exceptionally high-density team of world-renowned computer vision researchers, you will own core architectural and pre-training breakthroughs for their next generation of multimodal visual and video foundation models.

KEY RESPONSIBILITIES
• Conceive, design, and pre-train multi-billion-parameter rectified-flow and diffusion-transformer (DiT / MMDiT) architectures unifying text-to-image, instruction-guided editing, in-context visual reasoning, and temporal video synthesis.
• Innovate on continuous and discrete visual/spatiotemporal tokenizers (autoencoders) maximizing perceptual reconstruction fidelity, compression ratio, and latent-space trainability.
• Design post-training alignment recipes—spanning DPO, differentiable reward backpropagation, and RLHF for aesthetics, physical plausibility, text rendering, and multi-subject compositionality.
• Scale stable mixed-precision pre-training runs across thousands of GPUs, diagnosing gradient spikes, representation collapse, and multi-resolution curriculum tradeoffs.
• Curate and filter billion-scale image-video-caption training mixtures using synthetic VLM recaptioning and semantic deduplication.
• Publish landmark technical reports and open-weight milestones while collaborating directly with inference engineers to bring frontier models to millions of users.

TECHNICAL ENVIRONMENT & OFFERING
Researchers enjoy massive dedicated H100/B200 compute per scientist, zero bureaucratic overhead, and a clean PyTorch/CUDA distributed training codebase. Offered hybrid in Freiburg or Berlin, compensation includes €125,000–€180,000 EUR base salary plus premier frontier-lab equity, 30 days paid leave, and full relocation/visa support.

## Core Requirements

ESSENTIAL QUALIFICATIONS
• PhD in Computer Vision, Machine Learning, or a related field, or an equivalent track record of seminal industrial research in generative visual modeling.
• First-author publications at premier venues (CVPR, ICCV, ECCV, NeurIPS, ICML, ICLR, or SIGGRAPH) in diffusion models, flow matching, autoregressive visual generation, or neural tokenizers.
• Expert hands-on mastery of PyTorch and large-scale distributed multi-node GPU pre-training (FSDP, tensor/sequence parallelism, BF16/FP8).
• Strong intuition for scaling laws, multimodal text-vision conditioning, VAE/GAN latent representations, and perceptual evaluation metrics.
• Self-directed builder mentality—comfortable writing training loops, data loaders, and custom losses directly from scratch.

PREFERRED / BONUS EXPERIENCE
• Prior experience co-leading a frontier open-weight or commercial text-to-image / text-to-video foundation model release.
• Familiarity with custom CUDA/Triton attention kernels and step-distillation techniques.

## 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-research-scientist-flow-matching-visual or via HTTP `POST https://jobs.talent247.co/api/public/jobs/senior-research-scientist-flow-matching-visual/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-research-scientist-flow-matching-visual/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`.
