Forward Deployed Engineer

About H2Loop

H2LooP.ai is building next-generation sovereign coding models and context platforms for system engineers across embedded systems, robotics, UAV, avionics, automotive, consumer electronics, and industrial IoT.

Our core team comes from Google, Toshiba, Cisco, Bosch, Philips, and NXP.

Our enterprise customers operate in highly regulated, infrastructure-controlled environments — defense, aerospace, automotive Tier-1s, and industrial OEMs. Some run fully air-gapped networks with no internet egress; others run connected on-prem clusters, private clouds, or hybrid setups with strict egress filtering. All of them expect the platform to run on their infrastructure, under their security posture, with no assumptions about cloud connectivity. You will be the person who makes that work.

What you'll own

On-prem deployments — connected, restricted, and air-gapped

• Own the full deployment lifecycle for enterprise on-prem installs across the full connectivity spectrum: connected private cloud, egress-filtered datacenter, and fully air-gapped networks

• For air-gapped environments: build and maintain a self-contained deployment bundle — all container images, Helm charts, model weights, and runtime dependencies pre-packaged and transferable via secure media with no outbound internet required

• For connected and restricted environments: manage proxy/firewall configs, private registry mirroring, and egress allowlist requirements; know which H2Loop services call out and what for

• Adapt deployment configs for customer-specific constraints: OS baseline (RHEL, Rocky Linux, CentOS Stream, Ubuntu LTS), Kubernetes distribution (OpenShift, Rancher, vanilla K8s), compute substrate (VMware, bare metal, private cloud)
  

Customer environment scoping

• Conduct pre-deployment technical discovery: available hardware (GPU nodes, CPU-only fallback), OS baseline, Kubernetes distribution, network topology (connected, egress-filtered, or air-gapped), proxy/firewall rules, internal DNS setup, and identity/auth stack (LDAP, Active Directory, SAML, PKI)

• Produce a site-specific deployment plan and hardware readiness checklist before each engagement, calibrated to the customer's connectivity model

• Identify blockers early — under-provisioned GPU memory, missing kernel modules, incompatible container runtimes, missing egress exceptions — and negotiate workarounds with customer IT teams
  

Deployment tooling

• Build and maintain a scripted, reproducible installation workflow: Helm values overlays per customer, pre-flight validation scripts, post-install smoke tests

• Maintain a private registry mirror and model weight bundle that works in both connected-restricted and fully air-gapped environments — versioned, signed, and not reliant on public infrastructure

• Write clear runbooks for each supported deployment topology: connected on-prem, egress-filtered, and air-gapped
  

Customer support & escalation

• Be the primary technical contact for enterprise customers during deployment and early production

• Debug live issues across the stack — Kubernetes scheduling failures, GPU driver mismatches, TLS misconfiguration, database connectivity, LDAP bind failures, model inference errors

• Know when to escalate to platform engineering and how to capture the information needed to do so efficiently
  

Feedback loop to engineering

• Translate field deployment pain into concrete product improvements: packaging gaps, missing configuration surfaces, undocumented environment assumptions

• Maintain a field issues log that the platform team can act on
  

Required experience

• 5+ years deploying complex software in enterprise or regulated environments — on-prem Kubernetes, private cloud, bare metal, or hybrid

• Hands-on experience across multiple on-prem deployment models: connected private datacenter (proxy/egress config, internal DNS, private registries) and air-gapped environments (pre-pulled images, offline Helm repos, OS package mirrors, artifact transfer via removable media)

• Kubernetes operations — namespaces, resource quotas, persistent volumes, ingress, RBAC, service accounts; comfortable reading and editing Helm charts

• Linux fundamentals at depth — systemd, kernel modules (especially GPU drivers: NVIDIA CUDA, ROCm), storage configuration, network debugging (tcpdump, ss, nftables/iptables)

• Enterprise auth integration — LDAP/Active Directory bind configuration, SAML SP setup, TLS cert chains, PKI basics

• Comfortable working on customer sites with limited tooling: no package manager internet access, restricted shell environments, change-control windows
  

Strong-to-have

• OpenShift or Rancher deployment experience — OLM, SCCs, route vs. ingress differences

• NVIDIA GPU stack depth — driver version compatibility, CUDA container toolkit, MIG configuration, GPU operator

• Experience deploying LLM inference stacks (vLLM, Ollama, TGI, or similar) on-prem — model weight management, quantization tradeoffs for constrained hardware

• Familiarity with defense or aerospace IT environments — DISA STIGs, RMF, ITAR-adjacent data handling, TEMPEST awareness

• Scripting for deployment automation — Bash, Python, or Ansible

• Container image hygiene — multi-arch builds, image signing (cosign), SBOM generation, vulnerability scanning for air-gapped compliance requirements
  

What we don't need

Someone who spins up a cloud environment for every problem. Our customers range from enterprise datacenters with strict egress policies to fully air-gapped defense networks. You need to meet each of them where they are — read the environment accurately, adapt the deployment to fit their constraints, and ship a working system without asking for exceptions you won't get.

First 90 days

1. Days 1–30: Shadow platform and DevOps engineers to fully understand the H2Loop deployment architecture — services, dependencies, GPU inference stack, and current on-prem packaging. Do a dry-run deployment into the internal air-gap test environment.

2. Days 30–60: Participate in one live enterprise deployment end-to-end. Identify gaps in current packaging and runbooks. Begin building automated pre-flight validation tooling.

3. Days 60–90: Own an enterprise deployment from kickoff to handoff independently. Deliver a structured field report to platform engineering with actionable packaging improvements.
   

Compensation & logistics

• Competitive early-stage equity + salary

• In-person, Bangalore office; travel to customer sites required (India and international)

• Small team — direct access to founders and engineering