Work with Amrik Singh Khalsa

Services, each backed by shipped work

Six ways to bring him in. Each one maps to work he has already shipped; the proof line under it comes straight from his resume.

1. 01

   AI agent architecture and design review

   An independent review of an agent system before it ships or scales: orchestration, tool boundaries, identity and credentials, failure handling, evaluation and cost. You get a written assessment and prioritized recommendations.

   For

   Teams with an agent prototype heading to production, or a production agent that has become hard to trust or explain.

   Proof

   Published 80+ architecture and research documents at Okta (Auth0) and delivered 6 agent-platform proofs of concept end to end.

2. 02

   Multi-agent systems, MCP servers and A2A integration

   Design and build of orchestrator and subagent fleets, MCP servers that expose your systems as agent tools, and A2A protocol integration between agents, with output that cites its evidence and says plainly what it could not reach.

   For

   Product and platform teams connecting agents to internal systems, or moving from one agent to several that cooperate.

   Proof

   Architected an orchestrator, analyst and synthesizer fleet with 4 parallel subagents that root-caused 11+ production escalations, and an A2A protocol runtime with 1,200+ tests green.

3. 03

   AI agent security

   Security gates for agents and executable skills: CI checks before a skill is distributed, prompt-injection defense, secret redaction in tool output, tamper-evident audit logs, and short-lived credentials through GitHub OIDC instead of long-lived secrets.

   For

   Organizations letting teams publish or run agent skills, plugins and MCP tools, and the security teams asked to approve them.

   Proof

   Owns the CI security gates for an org-wide marketplace of 170+ executable agent skills at Okta (Auth0), and merged secret redaction and audit-log fixes into the Microsoft Agent Governance Toolkit.

4. 04

   RAG and knowledge agents

   Retrieval-augmented agents over your own documents: ingestion pipelines, pgvector search, answers grounded in cited sources, and evaluation that shows whether retrieval is actually helping.

   For

   Support, documentation and internal-tools teams sitting on a large body of knowledge that people struggle to search.

   Proof

   Architected a GenAI support agent on RAG and pgvector with automated ingestion of 2,000+ technical articles, cutting internal query resolution time by 40%.

5. 05

   Fractional AI architect or founding-engineer tech lead

   Hands-on technical leadership for an early-stage team: architecture decisions, the first production systems, security foundations and a delivery rhythm, working alongside the founders.

   For

   Startups that need senior technical leadership on AI or platform work before, or instead of, a full-time hire.

   Proof

   As Founding Engineer and Tech Lead at Incentiv, led the team behind the core platform: $5M+ in ESOP liquidity, with 30+ startups and 200+ investors onboarded in 6 months.

6. 06

   Full-stack product engineering

   Production web products in TypeScript: Next.js frontends, NestJS and Prisma service layers, PostgreSQL, and authentication with OAuth2, JWT and RBAC, including the AI features inside them.

   For

   Teams that need a senior engineer to ship a product or a feature end to end.

   Proof

   At Incentiv, NestJS and Prisma service layers cut response times by 25% and Next.js UIs loaded 35% faster; at Augmento Labs, a component library served 6+ micro-frontends.

Four ways to engage

Pick the shape that matches the problem. Scope, deliverables and terms are agreed in writing before anything starts.

1. 01

   Scoped review

   A focused look at one system, design or decision, with the scope agreed up front.

   You get

   • A written assessment of what holds up and what does not
   • Prioritized recommendations you can act on
   • A walkthrough session with your team

   Fits when You want a second opinion before you build, ship or scale.

2. 02

   Build engagement

   Hands-on design and delivery of a defined piece of work, as a freelancer or embedded in your team.

   You get

   • Working code in your repositories, with tests
   • Architecture notes and decision records
   • Documentation and a handover, so your team owns the result

   Fits when The work is clear enough to scope, and you want it built.

3. 03

   Ongoing advisory

   Recurring architecture and design input while your team builds.

   You get

   • Reviews of designs and pull requests
   • Written recommendations on decisions as they come up
   • A sounding board for your tech lead or founders

   Fits when Your team is building and wants senior input without a full-time hire.

4. 04

   Full-time or remote role

   A permanent role as a senior engineer, software architect or tech lead on AI agent systems.

   You get

   • Ownership of agent architecture, security and delivery
   • Hands-on code alongside technical leadership
   • Five years of building and leading full-stack platforms

   Fits when You are hiring for a full-time role, fully remote roles included.

How an engagement runs

1. 01

   Brief

   Send a short brief through the form below or by email. A few paragraphs on what you are building and where you are stuck is enough.

2. 02

   Intro call

   A conversation about the context, the constraints and whether this is a good fit. If it is not, you will hear that directly.

3. 03

   Written scope

   Scope, deliverables, engagement model and terms are agreed in writing before any work starts.

4. 04

   Work and handover

   The work happens in your repositories and tools, with written updates along the way, and ends with documentation your team can own.

Questions about working together

• What kind of projects does Amrik Singh Khalsa take on?

  Amrik Singh Khalsa takes on AI agent work: architecture and design reviews, multi-agent systems, MCP servers and A2A integration, AI agent security, and RAG knowledge agents. He also works as a fractional AI architect or founding-engineer tech lead for startups, and on full-stack products in Next.js, NestJS and TypeScript. Every engagement is scoped in writing first.

• Does Amrik Singh Khalsa work with teams outside India?

  Yes. Amrik Singh Khalsa is based in Bangalore, India, on India Standard Time (UTC+5:30), and is open to working remotely with teams anywhere. Engagements run over video calls, email and your own tools, such as your repositories and issue tracker, so your team does not need to be in India or in his time zone.

• How do I start working with Amrik Singh Khalsa?

  Send a short brief through the form on this page, or email singhamrikkhalsa@gmail.com. Describe what you are building, the problem you need solved and any constraints. If there is a fit, the next step is an intro call, and then a written scope covering deliverables, engagement model and terms before any work begins.

• Is Amrik Singh Khalsa open to full-time or remote roles?

  Yes. Alongside consulting and freelance work, Amrik Singh Khalsa is open to full-time roles, including fully remote roles, as a senior engineer, software architect or tech lead on AI agent systems. His resume covers his work at Okta (Auth0), Incentiv and Augmento Labs, and his merged open source contributions at NVIDIA, Microsoft and Warp.

• How is a consulting engagement with Amrik Singh Khalsa priced?

  Pricing depends on the engagement model (a scoped review, a build engagement or ongoing advisory) and on the scope, so there is no public rate card. The budget range in your brief helps him suggest the model that fits, and the price is agreed in the written scope before any work starts.

• What should a project brief include?

  A useful brief says what you are building and for whom, where it stands today, the problem or decision you need help with, and any deadlines or constraints. Links to code or documents he can read help too, and a rough budget range helps him suggest the right engagement model. A few paragraphs are enough; no formal spec is needed.

• What does an AI agent architecture review cover?

  A review looks at how an agent system is put together and where it could fail: orchestration and tool boundaries, identity and credentials, exposure to prompt injection, secret handling, audit logging, evaluation and cost. The result is a written assessment with prioritized recommendations, drawing on the agent security and multi-agent work Amrik Singh Khalsa does at Okta (Auth0).