Hi! I'm Malavika. I'm passionate about building, writing, and innovating.
About Me
I’ve always been interested in both the arts and sciences, and that blend of analytical and creative thinking naturally led me to product. I studied Intelligent Systems at Columbia University, graduating in 2020. I think each chapter of my career has sharpened how I think about building.
I started off as a Software Engineer at Prognos AI, building an AI application that predicts risk scores, which gave me a deep appreciation for what it takes to make models reliable in high-stakes contexts. I then joined Microsoft as a Product Manager, and am currently leading our multimodal platform strategy at Azure OpenAI/Foundry, focusing on image/video. I grew and expanded our portfolio into new modalities and partnerships. 1 year ago, our team launched Dalle, and now we’ve launched and scaled models from Sora 1, Sora 2, all ImageGen models, to Black Forest Labs.
I'm passionate about bringing nascent capabilities to market in ways that are both impactful and safe — and I'm drawn to organizations where those two goals are treated as inseparable.
Experience
Senior Product Manager, Azure OpenAI
Grew multimodal portfolio from Dalle to Sora, Sora 2, All ImageGen models, Black Forest Labs, and more.
Pioneered Video Generation API ahead of OpenAI, giving Azure OpenAI first mover advantage
Scaled GPT-image-1 to 300+ billion tokens
Product Manager II, Azure Edge + Platform
Awarded Azure Edge + Platform Impact Award
Designed (0->1) and expanded code analysis in the Pull Request, reducing time to deliver analysis feedback from 45-60d to 1h. Projected to save Microsoft $55 million+
Product Manager Intern, Visa
Produced a Go to Market Strategy for Visa Secure Remote Commerce (SRC)
Collaborated with a team of 3 engineers to build and test an application that utilizes raw data from health insurance companies rosters and an AI model to predict risk scores
Software Engineer Intern, Prognos AI
Personal Projects
GenAI-Research Agent: Built a research monitoring agent that fetches papers from arXiv across machine learning, computer vision, and AI — then uses Claude to score each one for relevance to generative AI on a 1–10 scale. Papers scoring 7+ get a deeper analysis: what capability they unlock, why it matters, and a two-sentence summary. Results are deduplicated in a local database and compiled into a weekly digest grouped by capability area (reasoning, image generation, efficient training, etc.).
Tools used: Claude Code
