jsaunders909.github.io/_data/cv.yml at master · jsaunders909/jsaunders909.github.io · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
- title: General Information
  type: map
  contents:
    - name: Full Name
      value: Jack R. Saunders
    - name: Location
      value: Chippenham, UK
    - name: Email
      value: jacksaunders909@gmail.com

- title: Education
  type: time_table
  contents:
    - title: PhD, Computer Science
      institution: University of Bath, Bath, UK
      year: 2019 - 2025
      description:
        - Investigated the interaction of style and speech in facial animation, focusing on capturing and reproducing style in 2D and 3D facial Avatars.
        - "Techniques: differentiable rendering, neural rendering (NeRFs, Gaussian Splatting), Generative AI (diffusion models, transformers, GANs), inverse rendering, multimodal audio-visual AI."
        - Obtained a Master's with Distinction. Served on the Centre's Industry board, organised and chaired a conference.
        - ART-AI CDT, supervised by Vinay Namboodiri and Darren Cosker.
    - title: BSc, Mathematics — First Class Honours
      institution: University of Southampton, Southampton, UK
      year: 2016 - 2019
      description:
        - ET Davis Prize for highest average grade in pure mathematics.
        - Dean's List 2017, 2018 & 2019.

- title: Experience
  type: time_table
  contents:
    - title: Senior Research Scientist
      institution: Tavus, London (Hybrid), UK
      year: Feb 2025 - Present
      description:
        - Contributed to Phoenix-4, Tavus's animation and rendering model for real-time video generation of avatars, using Gaussian Splatting, Flow Matching, and GANs. Developed a novel implicit Gaussian Avatar model improving expressiveness, achieving a 25% increase in lip-sync metrics vs Phoenix-3.
        - Lead a research team investigating video diffusion models as a next-generation alternative to current avatar rendering pipelines, with a focus on pushing towards real-time capability.
        - "Key research directions: identity stability in autoregressive generation, Self-Forcing and related techniques for eliminating exposure bias, model distillation and quantization for real-time deployment."
        - Runs a significant portion of the hiring process — leading technical interviews, supervising intern projects, and mentoring junior researchers.
    - title: Founder & Freelance Consultant
      institution: RealSync AI, Chippenham, UK
      year: Apr 2024 - Present
      description:
        - Founded RealSync to provide consultancy and coding services for Avatar companies.
        - Helped several clients build prototypes now in production with tens to hundreds of thousands of users.
    - title: AI Research Science Intern
      institution: Microsoft, Redmond, WA, USA
      year: Jun 2024 - Sept 2024
      description:
        - Built a system for creating real-time, animatable, photorealistic Avatars from a single image or short video using Gaussian Splatting and synthetic data.
        - Led a research project with 10+ members (research scientists, engineers, designers, PMs); built a working prototype.
        - Work led to a CVPR 2025 paper (GASP) and a filed patent.
        - Also assisted on an LLM + scene graph spatial reasoning project, leading to a filed patent and a paper under review.
    - title: AI Scientist
      institution: DeepReel, London (Remote)
      year: Jun 2022 - Apr 2024
      description:
        - Early-stage employee at an AI-driven digital human company producing photorealistic custom avatars.
        - Proposed the pipeline for the prototype product using cutting-edge research methodology.
        - Proposed and won grant funding for a project improving the company's models.
        - Implemented 3D reconstruction, neural rendering, transformer, and CNN-based models in PyTorch.
    - title: Research Intern
      institution: Epic Games, Manchester (Remote)
      year: Nov 2021 - May 2022
      description:
        - Developed a novel style transfer algorithm for emotion changes in MetaHuman characters, presented at FMX 2022.
        - Contributed code to the Horizon 2020 project PRESENT.

- title: Publications
  type: time_table
  contents:
    - title: "GASP: Gaussian Avatars with Synthetic Priors"
      year: 2025
      description: "Saunders, J. et al. CVPR 2025."
    - title: "Dubbing for Extras: High-Quality Neural Rendering for Data Sparse Visual Dubbing"
      year: 2025
      description: "Saunders, J. & Namboodiri, V. BMVC 2025."
    - title: "TalkLoRA: Low-Rank Adaptation for Speech-Driven Animation"
      year: 2024
      description: "Saunders, J. & Namboodiri, V. BMVC 2024."
    - title: "FACTS: Facial Animation Creation Using the Transfer of Styles"
      year: 2024
      description: "Saunders, J.; Caulkin, S.; Namboodiri, V. Eurographics 2024."
    - title: "READ Avatars: Realistic Emotion-controllable Audio Driven Avatars"
      year: 2023
      description: "Saunders, J. & Namboodiri, V. BMVC 2023 (Oral)."

- title: Reviewing
  type: time_table
  contents:
    - year: 2025
      items:
        - "CVPR 2025, ICCV 2025, NeurIPS 2025, ICLR 2025"
    - year: 2024
      items:
        - "BMVC 2024 (Outstanding Reviewer), ECCV 2024, NeurIPS 2024"
    - year: 2023
      items:
        - "BMVC 2023"

- title: Honours & Awards
  type: time_table
  contents:
    - year: 2024
      items:
        - BMVC 2024 Outstanding Reviewer
    - year: 2019
      items:
        - ET Davis Prize — highest average grade in pure mathematics, University of Southampton
        - Dean's List (2017, 2018, 2019), University of Southampton

- title: Skills
  type: nested_list
  contents:
    - title: Programming & Frameworks
      items:
        - Python, PyTorch
        - Neural Rendering (Neural Textures, Gaussian Splatting, NeRF)
        - Generative AI (Transformers, GANs, Diffusion Models, Flow Matching, Video Diffusion)
        - Real-time model optimisation (Self-Forcing, distillation, quantization)
        - Inverse Rendering, Optimisation
    - title: Research & Communication
      items:
        - Research, Paper Writing, Grant Writing
        - Technical Interviewing, Mentoring, Team Leadership

- title: Interests
  type: list
  contents:
    - Football/Soccer, Rugby, Running, Gardening, Cooking