Miju Labs

← All buyers · Pika

C

Chenlin Meng

Co-founder & CTO · Pika

low confidenceTier B · Map and watchImage, video & audio modelsvideo

Meng is Pika's co-founder and CTO and the stronger research name of the pair — her Stanford work with Stefano Ermon includes SDEdit, a widely used image-editing method, and contributions to consistency and distillation models. She is the person at Pika who would specify data if Pika decided to compete on model quality; there is no public evidence it has.

Background

Meng did her PhD at Stanford in Stefano Ermon's group, the same lab that produced Jiaming Song, working on diffusion and score-based generative modelling. SDEdit (guided image synthesis and editing with stochastic differential equations) is her best-known paper and is still used in editing pipelines across the industry.

She left to co-found Pika with Demi Guo in 2023. Pika's public engineering communication is minimal — no technical reports, no evaluation reports, no research blog of note — so her current work is largely invisible.

What they run now

Pika's model and infrastructure stack, inferred. No public detail on training data, post-training, preference optimisation or evaluation.

What they buy

Unknown, likely little. By inference from her research lineage she would care about preference data if Pika ran a post-training programme, but nothing public indicates one exists.

Career

  1. 2023 – presentCo-founder & CTO, Pika
  2. c. 2019 – 2023PhD, computer science (Ermon group), Stanford University — SDEdit; consistency/distillation work
  3. c. 2019–2023Stanford University, PhD, computer science
Why it matters here

Low now, worth a watch item. She has the research background to build a proper post-training and preference pipeline, and if Pika ever moves upmarket she is where that starts. She is also a plausible warm-intro node into the wider Ermon-lab network, which includes Jiaming Song at Luma.

How to reach

No outreach recommended now. Speculation: her real value to Julian is as a network node rather than a buyer — the Stanford Ermon diffusion cohort is scattered across Luma, Pika and elsewhere, and being known in it is worth more than a Pika pitch.

What we could not establish
  • Whether Pika has any post-training or human-preference programme.
  • Her current technical focus — no public material found.
  • No confirmed public handles; the Scholar profile ID should be verified before use.

Why they are on the map

CTO and research lead (diffusion researcher); the post-training / preference-study hire would likely report to her. source

Needs the data for a model; writes the spec.

How Pika buys data

2026 Ashby role 'Research Scientist, Post-Training — Video Generation': 'Lead post-training evaluation, including human preference studies', 'design/configure preference data collection workflows, train and validate learned judges'. A separate 'Research Scientist, Data' role covers labeling, filtering and QA tooling. No vendor named; last disclosed funding was the $80M Series B (June 2024).

Founder-led: CEO Demi Guo and CTO Chenlin Meng, who own research. Small research team in Palo Alto with Post-Training, Data and Inference hires open. The creative director and brand leads shape the look.

Sources

  1. Pika: Demi Guo and Chenlin Meng
  2. Pika — Forbes profile