Sillah Babar

Deployed Kestra as a pipeline orchestrator on a Kubernetes cluster, configured fault-tolerant data extraction jobs on pre-emptible pods, and integrated Pub/Sub for asynchronous messaging. Built scalable AI pipelines using embeddings for vendor and product recommendations, developed a ChatGPT deep search connector, and led development of internal tools to generate RFP signals and automate sales outreach. Created AI-driven features for leads recommendations, built data pipelines using GPT-4.1, Claude, and DeepSeek to extract structured data from spend reports, and fine-tuned LLMs for domain-specific extraction tasks. Implemented CI/CD with Grafana dashboards and designed a graph-based knowledge system for semantic document retrieval.

Led development of an internal tool to track sequence complexity, annotation counts, and QA performance for a team of 200 annotators. Extracted data and KPI metrics in C++ for model evaluation in autonomous driving projects for Audi AG, Cariad and AAI. Supported annotation spec improvements and applied AI techniques such as the Hungarian algorithm for assignment tasks. Maintained Python post-processing pipelines for LiDAR data processing and assisted cloud migration to AWS with Kubernetes, job DSLs, and Jenkins for 120+ pipelines.

Developed chatbots using LangChain and RAG, deployed across Cloud Run and VM. Built customer support bots for banking and a medical device regulatory chatbot, and implemented MLOps practices for data versioning with DVC and MLflow. Implemented CI/CD with Docker and Jenkins.