Divit Mallah

I’m an AI Engineer focused on turning real-world data into reliable, user-facing intelligence. Currently, I’m building production NestJS and TypeScript services that power AI analytics for wearable ingestion and personalized sleep features.

In my role, I developed LangChain workflows using Gemini, structured outputs, and health-data tools to generate evidence-backed insights. I integrated the Gemini Files API to ground chart explanations in uploaded sleep and diet research papers, and I built Apple Health and Oura pipelines to convert biometric data into sleep and recovery metrics.

I also work at the systems level—improving reliability with caching, cron jobs, MongoDB optimization, and PM2 monitoring. Beyond that, I contributed to Raspberry Pi sensor and audio pipelines to support device automation and IoT communication, and I drive daily sleep-quality prediction using time-series metrics and pre-sleep behavioral signals.

Earlier, I interned as a Data Scientist at HERE Technologies, where I built Python and DuckDB pipelines for 100M+ mobility records and applied HDBSCAN to GPS trajectories to uncover mobility clusters and traffic patterns. I’ve also built geospatial visualizations, implemented OpenCV rules for object detection, and strengthened security log pipelines during my BARC trainee work with Winlogbeat, Logstash, and Elasticsearch. I’m motivated by building practical AI systems—whether that’s LLM explanation pipelines, federated learning, or deep-learning flood mapping—and by communicating impact through leadership roles like Director at WACA Foundation and editor/contributor at Freelunch.