Research Intern
Location: Gurgaon
Term: 3-6 Months
We are a venture-backed, stealth-stage technology company building next-generation matchmaking and relationship platforms. Our mission is to reimagine how people connect, using AI, community, and content as our building blocks.
We're not building another dating app — we're creating an experience where users feel: "This app gets me."
At the core of our product is a real-time, ML recommendation engine — similar to Spotify for song moods or TikTok for discovery.
Most dating apps struggle to recommend the right match because they simply don't know enough about you. When they try to gather more data, they face a tradeoff - fewer users stick around, and the pool shrinks. Swipe fatigue is real. We've found a novel way to break this loop to deliver deeply personalised recommendations - even in sparse data scenarios.
About the Role
We're hiring a research intern to work on cutting-edge generative vision problems, focused on identity + body preservation in generated images. This role is for someone who enjoys reading papers, building algorithms, training models, and validating improvements with strong experiments.
What You'll Do
- •Deep-dive into recent literature on Diffusion / FLUX, identity conditioning, and structure control
- •Build algorithms and prototypes to improve face identity preservation (reduce drift, improve likeness)
- •Improve body consistency (shape/proportions, pose stability, reduce warping/artifacts)
- •Train / fine-tune models and run structured experiments (baselines, ablations, comparisons)
What We're Looking For
- •Strong academics + research experience
- •Prior work with Diffusion / FLUX models (projects, papers, OSS, research work, or internship)
- •Familiarity with Diffusers / ComfyUI / SDXL, face embeddings & similarity metrics, and pose/segmentation/depth pipelines
- •Experiment-first mindset: can validate ideas with clean baselines and clear results
Why You'll Love It
- •Real ownership from Day 1 with visible impact on product quality
- •High learning curve: research → build → evaluate → iterate