Hi, I'm Sai .
I work on generative photo and video models at Canva 1 Senior Applied Scientist , Video Studio. Before that, Photo Effects, where I worked on Background Generator (1M+ monthly users)., and outside of that I like taking things apart to see why they work: why anything associative can be split in half , how every trainer in Pokémon Red would rank against each other, what's actually sitting inside Yakuza 0's data tables .
This site is where I write it down. Some of it is maths, some is ML systems, and a lot of it is advice for students trying to get into Australian tech 2 First Class Honours , Applied Data Science at Monash; thesis on bias in diffusion models. Once president of the Monash Association of Coding., written from not that long ago.
Recently
- A GPU-performance reading map for making deep learning faster: hardware hierarchy, rooflines, kernels, profiling, distributed training, serving, quantization, and monokernels.
- Exploiting associativity on binary operators to speed up operations
- Decoding Yakuza 0's data tables
- Framing recursion and dynamic programming as mathematical induction, with worked examples from linked lists and binary trees.
- A summary of some software I've been working on, and other updates.
Currently: probably napping.
Trainer card
Trainer Card No. 025
Sai Kumar M.K.
Senior Applied Scientist · Canva
MLHPCResearchMathsReverse eng
- Posts
- 27
- Dex
- 15
- Badges
- 4
- Since
- 2021
Pick a badge.
Project dex · highlights
No. 001
monash-handbook-plus
Unit search, cost calculator, areas of study browser, an interactive prerequisite graph, a degree planner with auto-scheduling, a pathway finder and shareable study plans.
No. 002
saikumarmk.com
This site: writing, the Yggdrasil skill tree, the project dex and a few 3D visualisations.
No. 003
mini-melbourne-3d
A new Mini Melbourne: buses, trams and V/Line in 3D.
No. 004
pokered-trainer-tournament
A recreation of the Pokémon Red Elo tournament on the pkmn engine.
Experience
Currently Senior Applied Scientist at Canva .
- Built a video-agent evaluation platform spanning 104 tasks across seven suites, with multi-turn checks, LLM-based scoring, tracing, and a results viewer; enabled engineers and partner research teams to inspect changes and evaluate models independently.
- Compared automated judgments with human reviews to identify gaps in creative-quality assessment; supported release evaluations and lower-cost model experiments, including separating narrative planning from execution.
- Implemented and tested transcription serving changes using vLLM and integrated forced alignment; helped reduce processing time for an hour-long video from approximately five minutes to one minute while improving recognition quality and word timing.
- Reduced LLM cost by 46% on a dialog evaluation by caching system prompts and tool definitions; deployed the caching integration to production.
- Profiled document deserialization and removed repeated stack inspection, reducing validation time from 1,484 ms to 1.3 ms on the measured workload.
- Developed detail-preserving Background Generation methods to restore high-frequency image detail for a widely used Canva feature; named inventor on the resulting filed patent application.
- Shipped a distilled-SDXL upgrade for Magic Expand before a release freeze, reducing end-to-end latency by approximately 50% through fewer denoising steps and GPU-resident image processing; evaluations showed fewer hallucinated people.
- Adapted a SLURM/
torchruntraining stack to Ray/Anyscale, replacing local-data assumptions with an S3-streaming loader and validating distributed training with platform and research teams.
- Developed and evaluated bias-mitigation methods for Stable Diffusion in Canva's text-to-image product, focusing on representation across gender and ethnicity.
- Presented internship research as part of Generation Gap: Addressing Bias in Generative AI, SIGGRAPH Asia 2023.
- Built a genetic-programming and simulation framework in Julia to investigate a game-theoretic problem in scientific publishing.
- Analyzed commit data from more than 30,000 deep-learning repositories using Python and the GitHub API to study the evolution of ML libraries for AutoML.
- First Class Honours (final honours grade: 90); Dean's List for academic excellence in 2021, 2022, and 2023.
- Thesis: Bias Modelling and Mitigation in Diffusion Models.
The one-page version: printable CV .