I build low-latency distributed systems, autonomous cloud and datacenter operations platforms, and real-time robotics stacks. Currently SDE Intern at Nava (autonomous cloud/datacenter ops) and pursuing Mechanical Engineering at IIT Ropar.
SDE Intern
Nava • Autonomous Cloud & Datacenter Ops (Go, Python)
Research Intern
RBCCPS, IISc Bengaluru • FOCAS Lab (Dr. Pushpak Jagtap)
B.Tech Mechanical Engineering
Indian Institute of Technology (IIT) Ropar • GPA: 7.51/10
A functional virtual environment. Run standard commands like ls, cat, and cd with Tab-completion and command history.
Direct technical impact across autonomous cloud infrastructure, cyber-physical robotics research, and competitive aerial systems.
Bengaluru, Karnataka
Core intelligence platform for autonomous cloud and datacenter operations management – Go, Python, Distributed Systems
Engineering the intelligence platform for autonomous cloud and datacenter operations — deterministic retrieval, agentic reasoning, and GPU-optimal inference.
Bengaluru, Karnataka
Formal Control and Autonomous Systems (FOCAS) Lab – Dr. Pushpak Jagtap – C++, Python, ROS2
Built and validated a distributed real-time control stack for indoor multi-UAV coordination on a physical 3-drone swarm.
Rupnagar, Punjab
Cumulative GPA: 7.51 / 10.0 • Focus on Robotics, Control Systems & Embedded Systems
Undergraduate engineering curriculum emphasizing multi-rotor dynamics, state estimation, autonomous navigation, and high-performance computing.
IIT Patna, IIT Bombay, IIT Madras
Aerospace, Aerial Robotics & System Engineering Competitions
Represented IIT Ropar in competitive multi-disciplinary engineering challenges, securing national recognition in aerial robotics and computational modeling.
Engineering Background
Systems software engineer working across autonomous cloud platforms, GPU runtime memory enforcement, and provably safe drone swarm coordination.
I build systems software — the kind that sits close to the metal and has no room for failure. I care about the architectural decisions that don't show up in the diff but determine whether a system holds together under real-world load.
Currently, I'm an SDE Intern at Nava, building the intelligence platform for autonomous cloud and datacenter operations. My work centers on Atlas (topology-aware graph DB with hybrid Graph-RAG via Brain ingestors for AWS/Nava Cloud — deterministic retrieval ≥5s to <200 ms, token cost 0), Cortex (OpenCode-style agentic harness grounded in live Atlas context with typed tool registry across AWS/Nava Cloud/Sunbird DCIM/Optimality Engine), and the Optimality Engine(GPU-cluster simulator for vLLM/SGLang — ~10s/sim at >96% accuracy, orchestrating 1M+ config sweeps in <1 min), plus observability pipelines for topology-aware RCA and auto-remediation.
Previously, as a Research Intern at IISc Bengaluru (Robert Bosch Centre for Cyber-Physical Systems, FOCAS Lab under Dr. Pushpak Jagtap), I constructed and validated a distributed real-time control stack for indoor multi-UAV coordination with motion-capture-based localization on ROS2/Raspberry Pi via MAVLink (50Hz fusion, sub-5ms deterministic latency) and a CBF safety layer for dynamic geofencing and inter-agent collision avoidance with zero violations in time-varying environments.
I also study Mechanical Engineering at IIT Ropar('23–'27), where I serve as Secretary of the Aeromodelling Club and Deputy Contingent Leader for Inter IIT Tech Meet 14.0.
At Nava, built Atlas (hybrid Graph-RAG via Brain ingestors — ≥5s to <200 ms, token cost 0), Cortex (OpenCode-style agentic harness), and Optimality Engine (~10s/sim, >96% accuracy, sweeps 1M+ configs in <1 min).
At IISc Bengaluru (RBCCPS FOCAS Lab), built distributed real-time C++/ROS2 control stack with motion-capture localization (50Hz, sub-5ms) and CBFs for provably safe concurrent transitions with zero violations.
Built GPUMan in C++ over Unix Domain Sockets with NVML polling to enforce per-process VRAM limits without hardware MIG, preventing system-wide CUDA OOMs in shared compute environments.
Engineered qLPV proportional-integral observers for quadrotors detecting motor loss in ~417ms with geometric tracking on SE(3) relinquishing yaw for stable post-failure hover.
Systems & Robotics
Nalin Angrish
SDE Intern @ Nava • Ex-IISc RBCCPS • IIT Ropar '27
IIT Ropar
B.Tech Mechanical (GPA: 7.51/10)
Aeromodelling Secretary
Deputy Leader, Inter IIT 14.0
Technologies deployed across autonomous cloud/datacenter ops at Nava, research at IISc Bengaluru, and C++ systems.
Systems & Infrastructure: Linux runtime internals, low-level daemons, and GPU memory management.
Multi-threaded daemons, NVML polling, Unix domain sockets, Eigen3 vectorization
gRPC streaming pipelines, distributed services, concurrent goroutines (Nava)
Process lifetimes, signal handling (SIGTERM), POSIX sockets, systemd services
Per-process VRAM inspection, memory limit enforcement, GPU telemetry
Multi-stage images, container isolation, pod management, deployment specs
AWS (EC2, S3), GCP (GCE, GKE), cloud automation via Model Context Protocol
Practical implementations across AI cloud infrastructure, low-level compilers, multi-agent frameworks, and robotics.
February 2026 - March 2026Cloud Manager is a full-stack platform that uses LLMs to convert natural language into structured cloud operations with real-time tracking and execution logs.
December 2025 - January 2026GPUMan is a reliability tool designed for shared GPU environments, that prevents misbehaving processes from triggering a catastrophic, system-wide CUDA Out-of-Memory (OOM) failure.
June 2025 - July 2025FlexNN is a C++ POC Library, that can be used to develop and train fully connected neural networks with a flexible n-layer design and multiple activations.