AI Systems Engineer

Founder AI Systems Engineer — Office of the Founder
Company: Digital Trust Infrastructure India Limited (DTIIL)
Function: Office of the Founder / Technology
Location: Pune preferred; hybrid may be considered
Experience: Approximately 3–6 years, capability more important than tenure
The Role
DTIIL is building an AI-native operating environment in which a small, high-talent-density human team is amplified by specialized AI agents while consequential authority remains with accountable humans.
We are looking for a hands-on Founder AI Systems Engineer to build the first internal system: FA-001 — Founder Agent.
This is not an IT-support role and not a conventional chatbot-development role.
You will work closely with the Founder to build a secure intelligence and orchestration system that can organize institutional knowledge, retrieve prior decisions, track hypotheses and commitments, conduct research, coordinate specialized AI agents, integrate approved enterprise tools, and surface matters that genuinely require Founder attention.
The objective is to build infrastructure that makes human judgment substantially more leveraged without transferring consequential human authority to AI.
What You Will Build
The initial system will include:
* Building Founder Agent
* structured institutional knowledge repository
* decision, hypothesis, commitment and Founder-attention registers
* secure model gateway supporting multiple AI providers
* local/private model capability where appropriate
* retrieval and provenance architecture
* Research Agent integration
* Founder Red Team capability
* agent permissions and human-approval controls
* evaluation framework
* audit/evidence logging
* secure mobile access from iPhone and iPad
* controlled integrations with approved email, calendar, repositories and other enterprise systems
The system should be model-independent wherever practical.
Responsibilities
Design and implement the FA-001 architecture.
Build secure APIs and agent/tool integrations.
Create structured retrieval over DTIIL institutional knowledge.
Implement model routing across approved cloud and local models.
Develop agent workflows using tool/function calling and appropriate agent protocols.
Build human-authorization gates for consequential actions.
Implement authentication, authorization, secrets management and audit logging.
Develop evaluation suites that test factual grounding, retrieval, contradiction detection, permissions and agent behaviour.
Integrate Git/GitHub-based engineering workflows.
Deploy and operate local services, containers, databases and controlled cloud infrastructure.
Support local AI inference on Apple Silicon where appropriate.
Build a secure mobile-friendly interface/API for Founder access.
Document the architecture so the system is reproducible and does not depend on one engineer.
Continuously threat-model the system, including prompt injection, excessive permissions, data leakage, poisoned retrieval, model errors and unsafe tool execution.
Required Technical Capability
Strong Python.
Good TypeScript/Node.js.
Strong understanding of APIs, REST, JSON, webhooks and authentication.
Practical experience with modern LLM APIs.
Experience with tool/function calling and agentic workflows.
Good understanding of RAG, embeddings, retrieval and structured knowledge systems.
Git/GitHub.
PostgreSQL or equivalent database capability.
Docker/containerized environments.
macOS and Linux.
OAuth, IAM, permissions and secrets management.
Cloud fundamentals across AWS, Azure or GCP.
Software testing and observability.
Security-conscious engineering.
Highly Valuable
MCP experience.
Claude Code / Codex or comparable coding-agent experience.
Local/open-weight LLM deployment.
Apple MLX / Apple Silicon AI experience.
Vector databases.
Knowledge graphs.
Agent evaluations.
CI/CD.
Threat modelling.
Mobile/PWA development.
Enterprise integrations.
Experience building systems where provenance and auditability matter.
What We Are NOT Looking For
A candidate whose primary skill is prompt engineering.
A conventional desktop/IT-support engineer.
Someone who only knows how to assemble no-code AI workflows.
A researcher focused primarily on training foundation models.
Someone who treats model output as inherently reliable.
A candidate who cannot explain the security boundaries of an agentic system.
Engineering Philosophy
The system will operate according to:
Human Authority. Machine Capability. Verifiable Evidence.
AI may research, reason, retrieve, compare, draft, code, test and recommend.
Consequential corporate, architectural, regulatory, financial and production actions remain subject to appropriate human authority.
The engineer should be comfortable building systems where autonomy is deliberately bounded rather than maximized.
Ownership & Security
All code, repositories, cloud environments, credentials, infrastructure and documentation must remain under DTIIL-controlled accounts.
Production cryptographic private keys and unrelated sensitive systems will not be made available to the Founder Agent.
Access follows least-privilege principles.
Historical records must not be silently rewritten.
AI-generated conclusions must remain distinguishable from authoritative institutional decisions and evidence.
Success
Success is not measured by how impressive the chatbot appears.
Success means the Founder can securely ask:
“What genuinely requires my attention today?”
and receive a reliable, evidence-grounded answer based on current institutional knowledge, prior decisions, commitments, contradictory evidence and delegated agent work—with consequential actions returned to the Founder for authorization.

Work arrangement
No