Teams at KIEFER

Improve language model quality for Polish users by defining use cases, measuring model performance, building datasets and evaluations, fine-tuning or adapting NLP and LLM models, and implementing regression testing, monitoring, and production ML practices. The role also analyzes language errors, incorporates human review and market feedback, and develops reusable multilingual evaluation methods.

Develop and improve agent and long-context LLM capabilities through post-training, data generation, evaluation, agent design, and inference optimization. Build benchmarks, verifiers, diagnostic tools, and reliable multi-step workflows involving tool use, memory, retrieval, and error recovery. Collaborate with ML, Product, and Engineering teams to productionize successful experiments with regression checks and improved performance and cost.

Develop and improve automatic speech recognition models for real meeting audio. Responsibilities include training and fine-tuning models, preparing datasets, building benchmarks and regression tests, performing error analysis, optimizing inference, and deploying and monitoring production ML systems. The role collaborates with Engineering and Product to establish quality goals and prioritize improvements across accuracy, latency, reliability, throughput, and cost.