Applied AI & Systems Engineer, building sandboxed execution infrastructure, agent orchestration, and eval frameworks
Robotics & Automation engineer who moved into AI systems. I build the layer between LLMs and reliable production behavior: execution sandboxes, orchestration graphs, and evaluation harnesses that make agents debuggable instead of magical.
- Agent execution infrastructure at LastLab AI: hybrid sandboxed runtime with conditional routing between Kata Containers (GPU workloads) and Firecracker microVMs (CPU-only isolation)
- LangGraph-driven orchestration: stateful multi-step agent reasoning with typed tool contracts and recoverable execution graphs
- Eval frameworks: component-level golden-response testing for agent reliability
Agent reliability • Sandboxed execution • Specification-first systems design • Eval-driven development • The intersection of AI infrastructure and the physical world (robotics/UAV background)
B.Tech, Robotics & Automation (9.1 GPA) · Contributed to Gemini/Llama training pipelines @ Turing · Distributed systems engineering in Go @ Cubane · UAV systems architecture @ VimaanX
LangGraph • Firecracker microVMs • Agent evaluation • FastAPI • Distributed systems • UAV systems • Python production backends
Email: sakshambisen123@gmail.com
LinkedIn: linkedin.com/in/sbsaksham

