Tobias Lohse
Head of Engineering @Orca
Signup · Get unlimited contacts
WORK HISTORY
Head of Engineering @Orca
Orca DB is a memory-driven MLOps platform for predictive AI systems steered through interchangeable, auditable memory.Led 5 engineers building a retrieval-augmented classification system with millisecond inference and data-centric online learning through memory updates without retraining.Engineered pipeline for GPU-optimized fine-tuning of embedding models with custom contrastive losses, distributed gradient caching, and online hard negative mining.Prototyped PyTorch harness for shaping embedding-space decision boundaries with faster feedback, and fine-tuned LLMs for embedding-guided synthetic data generation.Built agent-native SDLC platform with eval harness and a compounding skill library.Unified release pipeline for cloud (GKE) and on-prem deployments with E2E gates.Acted as product owner, established career ladder, and evolved on-call processes.
EDUCATION
Freie Universität Berlin
Bachelor’s Degree, Double Major in Philosophy and Computer Science
evangelisches Gymnasium zum Grauen Kloster
High School, High School
Northwestern University
Bachelor Thesis, Theoretical and Mathematical Physics
University of Konstanz
Bachelor’s Degree, Physics
SKILLS
ABOUT TOBIAS LOHSE
I\'m a generalist engineering leader with 10+ years turning AI capabilities into production systems. I came to software through physics and philosophy, and I\'ve been shipping production code since 2012.I build across the whole stack — from ML research and backend architecture to design systems and the UIs that make everything usable. My recent work has been at the intersection of agentic AI and production engineering: fine-tuning models, shipping AI-native products, and building the agentic tooling that changes how engineering teams operate. I\'ve also built systems from scratch at startups, scaled enterprise-grade software through compliance regimes like SOC2, HIPAA, and IEC 62304, and led engineering teams from 5-person startups to 60-contributor orgs.I\'m a architect with a strong aversion to over-engineering. Code is cheap now and architectural judgment is the scarce skill — that\'s what I optimize for, whether I\'m writing the code myself or shaping how a team operates. I believe in distributing authority through clear objectives, and in evolving systems and processes iteratively, letting them improve through the friction of real use.What energizes me most is building AI systems that genuinely make a difference in people\'s work and lives — moving the frontier of what\'s possible while staying grounded in whether it actually helps.
This profile is compiled from publicly available professional sources. Unifers is not affiliated with or endorsed by LinkedIn. Request removal of this profile.