Sarosh Hussain / CTO & Partner, Pendoah

CTO and principal AI architect.
I build production AI and lead the teams that run it.

I design and build agentic AI, RAG and voice AI systems that run in production, and I lead the engineering and forward deployed teams that keep them running. I own the architecture and stay close to the code.

CTO & Partner at Pendoah. Hands-on AI engineering, forward deployed delivery and engineering leadership. Former enterprise CIO and PwC director.

Discuss an opportunity Explore selected work
Download resume
Sarosh Hussain, CTO and principal AI architect
Houston, TexasGlobal delivery experience
Years in technology and delivery
23+
AI/ML systems in production
30+
AI/ML systems built in the past three years
50+
Staff I lead in Pendoah’s software and AI group
20+
Technology department I oversaw as CIO, plus a 30+ extended team
45
Enterprises served across my career
100+

Leadership that stays close to delivery

Build the team.
Own the outcome.

As CTO and Partner at Pendoah, I oversee a 110-person firm and directly lead 20+ staff in its software and AI group. My work connects product strategy and enterprise roadmaps, customer research and UX, architecture and forward deployed delivery across hundreds of completed projects.

I also serve as fractional CTO for venture-stage companies, including two that have each crossed $1M ARR. Earlier, as CIO of a homebuilder of about 1,300 employees and about 2,500 contractors and suppliers, I oversaw a 45-person technology department plus a 30+ extended team, and I held director and practice leadership roles across enterprise consulting.

That combination matters when the same leader needs to develop people, make architecture decisions and resolve what is actually happening in production.

Explore my experience

Selected work

Hands-on work, in production.

All projects

Engineering + forward deployed delivery

Enterprise conversational AI

Led engineering and FDE delivery across agent execution, integrations and release operations. Used conversation and execution traces to separate agent failures from integration and workflow problems, then directed product and reliability work.

Designed and implemented a feature that shows the support AI where it falls short, so the team knows what to fix next. Earlier work shipped the fix for duplicate and conflicting customer replies when agent runs overlap and made sure every handoff actually reached a person.

Platform architecture + regulated delivery

AI analytics across 50 pharmacies

Architecture and delivery of natural-language-to-SQL analytics across 50 pharmacy locations for an approximately $100M enterprise.

Retrieved relevant database relationships before SQL generation, validated queries before execution, and linked user feedback to an administrator-reviewed correction loop.

Interim CTO + engineering recovery

A health platform back on track

Recovered a production API, stabilized the application estate and established engineering and release ownership during a three-month interim CTO engagement.

Rebuilt a reproducible baseline across mobile, API and agent services, resolved a Redis-compatible TLS fault, and used real-device authentication checks to support TestFlight beta delivery and technical handover.

Client work is presented without client names. The project library adds more client case studies, each from business problem to architecture and outcome.

What I build with

The stack in my production code.

On one support-AI platform rebuild, I authored 500+ pull requests in about eight weeks and reviewed and deployed about 300 more from the team. These are the tools in my own repositories and client systems.

Models and agents
OpenAI SDK · OpenAI Agents SDK · Anthropic SDK (Claude) · Azure OpenAI · AWS Bedrock · LangGraph · LangChain · Model Context Protocol (MCP)
Retrieval
pgvector · Qdrant · Azure AI Search · hybrid search and reranking
Voice AI
Azure OpenAI Realtime · Twilio · LiveKit
Durable execution and observability
BullMQ · Temporal · OpenTelemetry · Langfuse
Languages and platforms
Python / FastAPI · TypeScript / Node.js · Next.js / React · PostgreSQL · AWS · Azure

My product

ABACare, going to market.

Product case study

What the AI does

Outcomes for the people who use it.

All case studies

Most of my work runs where mistakes are expensive, in healthcare, legal and finance, with the checks that let a business trust it in production.

Open source

Open source for developers who work with AI agents.

GitHub

Open-source tools on npm

How I work

Make each deployment improve the next.

01

Set direction

Translate customer and business needs into product priorities, architecture decisions and clear team ownership.

02

Deliver with the team

Coach engineers and stay hands-on through implementation, integration, difficult incidents and production release.

03

Improve the system

Use conversation and trace analysis to find root causes. Turn recurring failures into shared capabilities and evidence-backed product improvements.

Work with me

The right scope. Clear ownership.

Book a 30-minute intro call (opens in a new tab)