Back to dashboard

video diffusion model, multimodal model with <5 years of experience

completed4 qualified1 runApr 22, 6:45 AMvideo-diffusion-model-multimodal-model-with-5-years-of-exper
Parsed2 topics · Junior · Hybrid
Generating seed nodes
0 proposed
Explored 0 queries
0/0 done
    3
    Expanding nodes
    queued
    4
    Qualifying candidates
    queued

    Qualified Candidates (4)

    HL

    Han Lin

    medium hireability

    Research Scientist Intern@Meta

    Previously: Research Scientist Intern @ Meta

    San Francisco, US

    • PhD student at UNC MURGe-Lab (advisor: Mohit Bansal) working directly on video diffusion and multimodal models
    • Key works: Ctrl-Adapter (ICLR 2025 Oral, adapts controls to any diffusion model), VideoDirectorGPT (multi-scene video generation via LLM planning), Dreamrunner (story-to-video), Bifrost-1 (bridging multimodal LLMs + diffusion), and training-free text-to-video guidance
    • Research Scientist Intern at Meta
    • Hireability: MEDIUM — 3rd/4th year PhD student (started ~2022-23), not immediately graduating but clearly industry-oriented with Meta internship and strong publication output (h-index 10). Prime post-PhD hire or potential early return
    XG

    Xiuye Gu

    medium hireability

    Research Software Engineer@DeepMind

    Previously: Research Software Engineer @ Google

    San Francisco, US

    • Co-author of VideoPoet (Best Paper ICML 2024, zero-shot video generation via LLM) and MAGVIT v2 tokenizer ('Language Model Beats Diffusion') — core contributions to video generation and multimodal models at Google DeepMind
    • MS Stanford, no PhD
    • Website explicitly states interest in video generation
    • Hireability: MEDIUM — settled at Google DeepMind with no recent job-seeking signals (no website updates in 180 days); position_update ~13 months ago likely an internal title change
    AG

    Agrim Gupta

    low hireability

    Research Scientist@Meta

    Previously: Research Scientist @ DeepMind

    San Francisco, US

    • Core video diffusion and multimodal researcher: co-authored VideoPoet (ICML 2024 Oral, 394 citations), Photorealistic Video Generation with Diffusion Models (ECCV 2024, 271 citations), MALT Diffusion for any-length video generation (2025), and Exploring Diffusion Transformer Designs (NeurIPS 2025 Oral)
    • Also multimodal work with VIMA
    • Recent Stanford PhD (2024, thesis: Generative Models of Vision and Action), ~2 years post-PhD — solidly within <5 years requirement
    • Hireability: LOW — pipeline signals show he moved from Google DeepMind to Meta ~3-4 months ago (LinkedIn scraped Jan 2026), likely still settling into new role
    DE

    Dave Epstein

    low hireability

    Member of Technical Staff@Anthropic

    Previously: Graduate Student Researcher @ University of California, Berkeley

    New York, US

    • Strong diffusion model researcher — 'Diffusion Self-Guidance for Controllable Image Generation' (NeurIPS 2023, 320 citations) plus video understanding work (CVPR 2020, ICCV 2021)
    • PhD from Berkeley (Efros lab) completed ~2024, now MTS at Anthropic pre-training team doing multimodal generative modeling
    • Directly matches query
    • Hireability: LOW — recently joined Anthropic (likely 2024), still settling into the role

    Runs

    #1completed0 qualified / 0 foundApr 22, 6:45 AM