Saltanat Tugayeva

ML Engineer Intern

Freedom AI Labs

Built and evaluated a multi-shot visual dubbing pipeline; reached 0.680 median StableSyncNet against a 0.705 commercial baseline while limiting edits to the speaker’s oral region.

• Developed Q-Dub, a production-oriented visual dubbing pipeline for film footage that synchronizes a speaker’s mouth to translated audio while preserving identity, head motion, expressions, and the original scene. • Built an end-to-end multi-shot pipeline combining shot detection, face tracking, active-speaker selection, VAD, audio-video alignment, generative lip synchronization, and source-preserving compositing; evaluated LatentSync, X-Dub, MuseTalk, InfiniteTalk, and other video-generation approaches.

Founder & AI Engineer

tryinsightengine.com

Built and launched an AI SaaS with a multi-agent content workflow; attracted 450+ registered users and was selected for the Top 25 Demo Day.

• Built and launched an AI SaaS in 7 weeks using Next.js, TypeScript, PostgreSQL, LangGraph, and Vercel. • Designed a multi-agent workflow with researcher, critic, and writer agents for generating social content in a user’s writing style. • Reached 450+ registered users and 5,000+ visitors from 85 countries, generated $125 in revenue, and was selected for the Top 25 Demo Day out of 77 projects.

Junior ML Engineer

SDU AI Lab

Built computer vision and multimodal ML prototypes, including a harassment-detection pipeline that achieved F1 0.80 on an internal benchmark.

• Built computer vision and multimodal ML prototypes, including a Kafka-based harassment-detection pipeline using EfficientNet-B2, LSTM, and OpenCV, achieving F1 0.80 on an internal benchmark.

B.S., Computer Science

Suleyman Demirel University (SDU)

Kairu × Astana Hub

nFactorial School