Haseeb Arshad

Conversational AI · enterprise integrations · production delivery

I’m a Forward Deployed AI Engineer at PolyAI, where I build the AI core of production voice agents: retrieval over customer knowledge bases, tool-calling functions, guardrails, and the evaluation that gates each release. I then own the path to production across SIP telephony, the real-time speech stack, and the enterprise systems each agent has to reach. Previously at PeerLogic, I designed multi-agent patient engagement and appointment workflows across voice and SMS. I specialize in the hard part after the demo: making an agent behave, integrate, and stay reliable once real callers are on the line.

Download CV Explore AI work

What I Do

Conversational AI

I build and deploy enterprise voice AI agents that handle real conversations — from appointment booking to lead capture — integrating telephony, EHR, and CRM systems into seamless automated workflows.

Back-End Engineering

I architect high-performance server-side systems using Python, Django, and Node.js — building scalable APIs, microservices, and data pipelines that power production applications.

AI / ML Engineering

I design and deploy LLM-powered solutions using LangChain, LangGraph, and RAG pipelines — from prompt engineering and multi-agent orchestration to production AI observability.

Platform Engineering

I build cloud-native platforms across AWS, GCP, and bare metal — Kubernetes, GitOps delivery, identity-aware access, replicated storage, and tested recovery.

Reliability & Observability

I connect user-visible availability to metrics, traces, logs, synthetic probes, capacity signals, and actionable alerts so teams can diagnose causes instead of guessing.

API Integration

I specialize in integrating complex third-party APIs — Salesforce, healthcare EHR systems, payment gateways, and telephony providers — building resilient, fault-tolerant data flows.

Professional AI work

Production AI that has to work in the real world

My primary work is shipping conversational AI for real customers: designing integrations, handling live-call constraints, validating releases, and staying accountable after launch.

View professional case studies

PolyAI

2025 - Present

Professional

Enterprise Voice AI and the LLM Core Behind It

I design and build the AI core of production voice agents: retrieval over customer knowledge bases, tool-calling GenAI functions, dialogue behaviour, guardrails, and evaluation, then own the path to production: enterprise integrations, SIP telephony, the real-time speech stack, and the monitoring that keeps a live programme honest.

Build RAG and tool-calling functions in Python, tuning semantic search and chunking over enterprise knowledge bases
Design dialogue behaviour, guardrails, and structured plus LLM-as-judge evaluation so releases cannot silently regress
Migrate legacy IVR flows to generative agents and connect CRM, EHR, scheduling, and customer-owned APIs behind resilient adapters
Own SIP telephony, ASR and TTS tuning, barge-in and latency work, and delivery through launch, monitoring, and hypercare

LLM AgentsRAGTool CallingPythonSIP TelephonyASR / TTSEvaluationKubernetes

Read the case study

PeerLogic

2022 - 2025

Professional

Multi-Channel AI for Patient Engagement

Engineered LLM-powered workflows across voice, SMS, practice systems, and human escalation for dental practices.

Designed multi-agent workflows with LangChain and LangGraph
Integrated NexHealth and major telephony providers through resilient adapters
Built Vue.js operations surfaces and Google Cloud Functions
Added observability and delivery automation for production releases

LangGraphLangChainDjangoVue.jsGCPWebhooks

Read the case study

Years in Software and AI

Supporting Lab Case Studies

Testimonials

"Haseeb is one of those engineers who just gets things done. He ramped up on our agent deployment stack incredibly fast, took full ownership of complex EHR integrations, and consistently delivered under tight go-live deadlines. A reliable teammate you can always count on."

Milos Kostic

Software Engineer, PolyAI

"Working with Haseeb on the Docs Dermatology project was a great experience. He took initiative on the API integration, built out the real-time config tooling, and was always proactive about catching issues before they hit production. He bridges the gap between engineering and product seamlessly."

Ryan Creeser

Product Support Manager, PolyAI

"Haseeb grew from a junior engineer to leading critical projects like AutoMarket Analytics and SecureTaxOffice. His ability to build scalable backend systems, mentor teammates, and take ownership of production infrastructure made him an invaluable part of our engineering team."

Ahmad Raza

Engineering Lead, Datics AI

Clients

Docs Dermatology

Healthcare

Grace Health

Healthcare

Azara Healthcare

Healthcare

Harmar

Home Accessibility

72Sold

Real Estate

SeekNow

Insurance

PeerLogic

Dental AI

AutoMarket Analytics

Automotive

DriveSmart CRM

Automotive

SecureTaxOffice

Government / Tax

Professional AI Experience

Building AI products that work in production

Forward-deployed conversational AI at PolyAI and LLM product engineering at PeerLogic, backed by 7+ years of full-stack and platform delivery.

Download Resume

2025 - Present

AI Experience Highlight

Forward Deployed AI Engineer

PolyAI

Voice AI, LLM Agents, RAG & Telephony

Design and build the AI core of production voice agents for enterprise contact centres: dialogue and tool-calling logic, retrieval-augmented context over customer knowledge bases, and guardrails that keep every response grounded and in scope
Tune retrieval quality with embedding-based semantic search and chunking strategy, and gate releases with structured and LLM-as-judge evaluation over real and synthetic call data
Migrate legacy intent-based IVR flows to generative agents and connect Salesforce, EHR, scheduling, Zoom, hospitality platforms, and customer-owned APIs behind adapters with explicit failure and escalation behaviour
Own the telephony path: SIP integration and migrations, call-failure diagnosis during onboarding, and hardened routing, transfers, and fallback for production traffic
Tune the real-time voice stack inside a live-call latency budget: ASR and TTS pipelines with custom brand voices, voice-activity detection and barge-in, audio caching, and runtime optimisation
Own delivery on AWS, GCP, and Kubernetes from sandbox validation through rollout, Datadog monitoring, conversation analysis over live call volume, and hypercare

2022 - 2025

AI Experience Highlight

Software Engineer

PeerLogic

LLM Products & Multi-Agent Workflows

Designed production LLM workflows with LangChain and LangGraph for conversation processing and downstream automation
Built multi-agent scheduling experiences across SMS and voice, including resilient orchestration around external systems
Delivered Vue.js interfaces, Google Cloud Functions, and integrations with GoTo, Bandwidth, NetSapiens, RingCentral, and NexHealth

2021 - 2024

Software Engineer

Datics AI

Delivered multiple client programs concurrently, including automotive analytics and public-sector tax platforms
Designed crawler infrastructure capable of coordinating up to 10,000 workers using Django, Scrapy, and Beautiful Soup
Created production AWS architectures, deployment pipelines, and database-backed integrations
Mentored engineers across back-end development, system design, and delivery practices

Aug 2020 - Nov 2020

Trainee Software Intern

Questra Digital

Delivered website features through a Git-based CI/CD workflow in a distributed team
Collaborated through Jira, Slack, and Trello to plan, review, and release changes

July 2019 - Oct 2019

Web Developer Intern

Petsaal Tech

Built and maintained customer-facing PHP, JavaScript, HTML, and CSS features

Education

2017 - 2021

University of Central Punjab

BSCS (4 years)

Core Courses

Artificial IntelligenceSoftware EngineeringIntroduction to Machine LearningProgramming FundamentalsComputer ArchitectureDatabaseOperating System

Engineering Courses

Human Computer InteractionSoftware TestingSoftware Quality AssuranceData Structure And AlgorithmsObject Oriented Analysis DesignMobile Application DevelopmentProgramming For Big Data

Technical Toolkit

AI Systems

Agents and retrieval

LangGraphLangChainRAGTool CallingModel Routing

Voice and language

Conversational AISIP/Telephonyfaster-whisperOllamaLLM Evaluation

Application Engineering

Back end

PythonDjangoDjango REST FrameworkNode.jsPostgreSQL

Front end and integrations

ReactVue.jsAstroOAuth 2.0WebhooksFHIR APIs

Cloud and Platform

Infrastructure

Kubernetesk3sAWSGCPDockerTerraformLinux

Delivery and reliability

Argo CDGitHub ActionsPrometheusGrafanaCaddyWireGuard

Licenses & Certifications

LangChain & Vector Databases in Production

ActiveLoop

November 2023

AWS Fundamentals: Going Cloud-Native

Coursera

August 2020

Complete Python/Django Mastery

Code with Mosh

September 2021

Complete SQL Mastery

Code with Mosh

September 2021

The Ultimate Docker Course

Code with Mosh

September 2021

The Ultimate Git Course

Code with Mosh

September 2021

Awards & Honors

Supporting platform depth

The infrastructure behind how I think about production

Outside my professional AI work, I operate a three-node Kubernetes environment to deepen the platform skills that support reliable AI delivery: immutable delivery, identity-aware access, replicated data, local AI, observability, backups, and tested recovery. The lab is supporting evidence, with 24 documented builds.

Explore every project

01Production

Core platform

Highly Available Kubernetes Homelab

A three-node, production-style k3s platform designed for node loss, repeatable recovery, and continuous GitOps reconciliation.

k3sKubernetesArgo CDMetalLB

control-plane nodes

embedded etcd

replicated state

02Production

Delivery engineering

GitOps Delivery and Immutable Releases

A source-to-production pipeline that builds containers, resolves immutable image digests, validates desired state, and lets Argo CD perform controlled rollouts.

GitHub ActionsGHCRArgo CDKubernetes

100%

digest-pinned images

website replicas

5 min

release detection

03Production

Zero-trust access

Identity-Aware Edge and Remote Access

A layered access plane combining automatic TLS, single sign-on, two-factor authentication, client certificates, and outbound-initiated VPN paths.

CaddyAutheliaWireGuardOAuth 2.0

edge replicas

2FA

operator access

mTLS

sensitive clients

Production AI field notes

Engineering AI systems that work in the real world

Practical lessons from shipping conversational AI, healthcare integrations, multi-agent orchestration, reliable telephony, and production observability.

See all articles →

API IntegrationJan 20, 2025

Integrating EHR Systems into Voice AI Workflows

Practical patterns for connecting Electronic Health Record APIs to real-time voice AI agents

AI EngineeringMar 15, 2025

Building Multi-Agent Conversational AI Systems with LangGraph

How to architect multi-agent systems for production conversational AI using LangGraph's graph-based orchestration

TelephonyNov 8, 2024

Production-Grade Telephony: SIP, Failover, and Reliability

Engineering reliable SIP-based telephony infrastructure for AI voice agents at scale

ObservabilitySep 12, 2024

Real-Time Monitoring for AI Agents with Datadog

Setting up comprehensive observability for conversational AI agents using Datadog's monitoring stack

Get in Touch

Contact Information

East Meadow, NY, USA

+1 516-757-9554

haseeb.arshad55@gmail.com

Remote Available