Saken Mukanov

AI Software Engineer

Perceptis

AI presentation-generation platform, 5-engineer team

Built production MCP integrations, GenAI routing and retrieval, and consolidated five services. Reduced image costs, release downtime, and review time.

• Built and launched an MCP server (async jobs, status tools, ETAs) for deck generation inside Claude and ChatGPT, listed in both directories; took it from gated beta to production in 17 days, now driving ~15% of all slide generation. • Built a multi-provider GenAI layer (Vertex AI, Bedrock, FAL) with structured outputs, fallback routing and per-session cost tracking with hard spend limits. Eval-driven model tiering cut image cost per slide by ~67% and total cost per slide by ~10–12%; a human-review eval caught the cheaper model inventing data, fixed and re-validated on ~100 slides.

AI/ML Engineer, National AI Platform

NITEC (National Information Technologies)

Automated model versioning and deployment for Kazakhstan’s National AI Platform, and built LLM, RAG, and NLP workflows for Kazakh legal documents.

• Built MLOps for Kazakhstan’s National AI Platform, automating model versioning and deployment; fine-tuned open-source LLMs and built RAG and NLP workflows for Kazakh-language legal documents.

Software Engineer, Data & Backend

Betty Data

Built an asynchronous pipeline covering 55 million YouTube videos and improved production ClickHouse query performance by 201%.

• Engineered an asynchronous pipeline over 55 million YouTube videos; improved production ClickHouse query performance by 201% through sharding, replication and data-layout optimization.

Software Engineer, Data & Backend

Rita Personal Data

Built ETL and backend workflows for 70,000+ users, migrated analytics to ClickHouse, and developed a customer behavior model with 0.71 F1.

• Built ETL and backend workflows serving 70,000+ users; migrated analytics from Snowflake to ClickHouse; developed a supervised customer-behavior model reaching 0.71 F1.

Software Engineer, Data & Backend

Alem School

Built software and data workflows for over one million daily COVID-19 metrics and mentored programming students for nine months.

• Built software and data workflows processing 1M+ daily COVID-19 metrics; mentored programming students for 9 months through code review, debugging and algorithmic problem solving.

M.Sc., Data Science

Nazarbayev University

B.Sc., Computer Science

Nazarbayev University