0%

About Experience Projects Skills Contact
AI Systems Architect

Saahith Matam

Building intelligent systems that think, learn, and scale — from autonomous AI agents to production conversational platforms.

Explore Work Let's Talk
Scroll
AI SYSTEMS• RAG PIPELINES• CONVERSATIONAL AI• FULL STACK• AGENTIC AI• LLM ENGINEERING• AI SYSTEMS• RAG PIPELINES• CONVERSATIONAL AI• FULL STACK• AGENTIC AI• LLM ENGINEERING•
01

The Architect
Behind the Intelligence

I'm an AI Systems Architect and Full Stack Engineering Leader with 5+ years of building systems that genuinely transform how businesses operate.

As the Founder and COO of StackForge LLC, I designed and built Percy — a production-grade autonomous AI agent capable of executing complex business operations end-to-end. Percy combines RAG pipelines, real-time web intelligence via FireCrawl and Perplexity, and multi-turn conversational reasoning into a single system that thinks and acts.

My career spans enterprise HR automation at Paycom, real-time conversational platforms at Vintatta, ML-powered analytics at IOpen Innovations, healthcare tech at Vitalus Health, and consumer social at HotMinute. I hold a B.S. in Computer Science from UT Dallas and an IBM certification in Python for Data Science & AI.

0+ Years
0+ Production Systems
0K+ Interactions
02

Where I've Built & Led

Mid 2024 — Apr 2026
AI Conv AI Cloud Leadership

Founder, COO & Senior AI Engineer

StackForge LLC Houston, TX

  • Founded and led StackForge LLC — an AI-first engineering firm specializing in intelligent automation and conversational AI for enterprise clients.
  • Architected Percy — a production-grade autonomous AI agent built on GCP, Gemini, Supabase, RAG, FireCrawl, and Perplexity API — executing complex business operations end-to-end without human intervention.
  • Engineered Percy's RAG pipeline with real-time web crawling and live knowledge retrieval, enabling deep domain-specific queries with high accuracy.
  • Built multi-turn conversational state management and intent orchestration on GCP for seamless context retention across long-horizon interactions.
  • Integrated Supabase as a hybrid data layer — vector embeddings for semantic search with relational storage for structured business data.
GCPGeminiSupabaseRAGFireCrawlPerplexityPythonTypeScript
Feb 2023 — Dec 2025
AI Conv AI Full Stack

Senior Software Engineer

Paycom Irving, TX

  • Designed conversational automation using NLP-based intent detection and contextual state management to modernize enterprise HR workflows at scale.
  • Engineered high-performance webhook microservices powering real-time conversational triggers and multi-step process orchestration.
  • Implemented async event-driven pipelines improving HR workflow efficiency by 30% across thousands of daily users.
  • Led full-stack tooling in React, TypeScript, and Python — improving engineering throughput and feature delivery.
ReactTypeScriptPythonNLPWebhooksMicroservices
Apr 2021 — Oct 2022
AI Conv AI Full Stack Cloud

Lead Full Stack Engineer

Vintatta LLC Houston, TX

  • Architected a real-time conversational platform processing 10,000+ monthly interactions with sub-100ms latency.
  • Designed NLP-driven intent routing with Azure Cognitive Services — improving classification accuracy by 35%.
  • Integrated OCR and Azure text analytics for intelligent document-based interactions with automated data extraction.
AzurePythonNLPOCRMicroservicesReact
Jan 2022 — May 2022
Full Stack Cloud AI

Lead Full Stack Engineer

IOpen Innovations LLC McKinney, TX

  • Led migration from legacy monolith to Django + PostgreSQL + GraphQL — improving uptime by 45%.
  • Built ML-driven analytics dashboards for enterprise data-driven decision-making.
  • Managed AWS infrastructure with Docker and Kubernetes for deployment reliability.
DjangoPostgreSQLGraphQLAWSDockerK8s
Apr 2021 — Oct 2021
Full Stack AI

Full Stack Engineer

Vitalus Health LLC Houston, TX

  • Built secure REST APIs in Node.js + PostgreSQL handling 1,000+ daily transactions for medical billing and scheduling.
  • Implemented NLP-powered form pre-fill reducing patient onboarding time by 40%.
Node.jsPostgreSQLReactNLPWCAG
Mar 2020 — Mar 2021
Full Stack AI Cloud

Full Stack Engineer

HotMinute LLC Houston, TX

  • Built a consumer social platform with React Native, Firebase, GraphQL — scaling to 5,000+ users in three months.
  • Deployed recommendation engine with user embeddings — increasing session duration by 22%.
React NativeFirebaseGraphQLPythonML
03

Selected Work

01
StackForge LLC — 2024–2026

Percy

Autonomous AI Business Operations Agent

Full-stack autonomous AI agent executing end-to-end business workflows — customer communication, scheduling, data retrieval, and decision support. Powered by Gemini, RAG, FireCrawl, Perplexity, and Supabase with dual-mode deployment.

GeminiRAGFireCrawlPerplexityGCPSupabase

How It Works

Percy operates as an autonomous business agent — given a task like "follow up with all leads from last week," it decomposes the goal into sub-tasks, retrieves context from Supabase, crawls relevant web data via FireCrawl, reasons through Gemini, and executes actions (emails, scheduling, data entry) without human intervention. Multi-turn conversational state keeps context across long-horizon workflows.

Key Skills Applied

  • RAG Pipelines — Retrieval-augmented generation for domain-specific knowledge with vector embeddings in Supabase
  • Gemini API — Multi-modal reasoning for task decomposition and natural language understanding
  • FireCrawl — Real-time web intelligence gathering for live data enrichment
  • GCP Cloud Run — Serverless deployment with auto-scaling for burst workloads
  • Perplexity API — Deep research queries for complex information synthesis
02
Personal — 2026

Sonara

Adaptive Music & Wearables Sync

Intelligent music platform synchronizing audio with biometric wearable data. AI-driven playlist generation adapting to heart rate, activity, and mood in real-time.

React NativeAI/MLWearable APIsPython

How It Works

Sonara connects to wearable devices (Apple Watch, Fitbit) via HealthKit/Wearable APIs to continuously monitor biometric signals — heart rate, step cadence, HRV, and activity type. An ML pipeline classifies the user's current state (resting, running, focused, stressed) and dynamically adjusts playlist generation through a recommendation engine trained on mood-activity-music correlations.

Key Skills Applied

  • React Native — Cross-platform mobile app with real-time biometric data streaming
  • Wearable APIs — HealthKit and Fitbit SDK integration for continuous sensor data
  • Python ML — Classification model for activity-mood state detection from biometric signals
  • Recommendation Engine — Collaborative filtering + content-based hybrid for playlist curation
Visit Live Site →
03
Sahasra Systems — 2025–Present

Sahasra

B2B SaaS Temple Management

Comprehensive platform for temple and religious org management — event scheduling, donation tracking, member management, role-based access, and real-time dashboards.

ReactNode.jsPostgreSQLSaaSCloud

How It Works

Sahasra is a multi-tenant B2B SaaS platform where each temple gets an isolated workspace. Admins manage events with automated reminders, track donations with receipt generation, maintain member directories with family groupings, and monitor engagement through real-time dashboards. Role-based access control ensures priests, treasurers, and committee members see only relevant data.

Key Skills Applied

  • React + Node.js — Full-stack architecture with SSR for fast initial loads and SEO
  • PostgreSQL — Multi-tenant schema design with row-level security for data isolation
  • SaaS Architecture — Subscription billing, tenant provisioning, and usage metering
  • Cloud Deployment — CI/CD pipeline with staging/production environments
Visit Live Site →
04
HackUTD — Oct 2021

AI Event Check-In

Real-time Face Recognition

AI-driven check-in for 1,000+ concurrent users with Google Vision AI, real-time face recognition, and session analytics.

FlaskGoogle Vision AIReactCV

How It Works

Attendees register by uploading a photo. At the event, a camera feed captures faces in real-time, sends frames to Google Vision AI for face detection, then matches against registered embeddings using cosine similarity. Check-in is confirmed in under 2 seconds with live analytics dashboards showing attendance rates, peak times, and session tracking.

Key Skills Applied

  • Google Vision AI — Face detection and embedding extraction for identity matching
  • Flask — Lightweight Python API handling real-time image processing pipeline
  • Computer Vision — Face encoding, cosine similarity matching, and confidence thresholds
  • React — Real-time dashboard with WebSocket updates for live check-in feed
05
Vintatta LLC — 2021–2022

Conv. Platform

Multi-Channel Bot Infra

10,000+ monthly interactions at sub-100ms latency. NLP intent routing with 35% accuracy gains and OCR-powered document interactions.

AzureNLPOCRMicroservices

How It Works

A unified conversational AI platform routing user intents across web chat, SMS, and voice channels. NLP models classify intent and extract entities, then a microservices orchestrator dispatches to the appropriate fulfillment service. OCR integration allows users to upload documents (invoices, IDs) and have them parsed and processed conversationally.

Key Skills Applied

  • Azure Bot Framework — Multi-channel deployment with unified conversation state
  • NLP / Intent Routing — Custom intent classification with 35% accuracy improvement over baseline
  • OCR Pipeline — Document parsing and entity extraction for structured data capture
  • Microservices — Event-driven architecture with sub-100ms response latency at scale
06
HotMinute LLC — 2020–2021

HotMinute

Social Platform + AI Recs

Consumer social app scaling to 5,000+ users in 3 months with ML recommendation engine increasing session duration by 22%.

React NativeFirebaseGraphQLML

How It Works

A consumer social platform built in React Native with Firebase backend. Users create time-limited content that expires after a set duration, driving urgency and engagement. A GraphQL API powers the feed with real-time subscriptions, while an ML recommendation engine uses user embeddings (interaction history, content preferences) to rank feed items and suggest connections.

Key Skills Applied

  • React Native — Cross-platform mobile app with smooth 60fps animations
  • Firebase — Real-time database, auth, cloud functions, and push notifications
  • GraphQL — Efficient data fetching with subscriptions for live feed updates
  • ML Recommendations — User embedding model increasing session duration by 22%
07
Vitalus Health — 2021

Healthcare Platform

NLP Medical Billing & Patient Mgmt

WCAG-compliant healthcare platform handling 1,000+ daily transactions. NLP-powered form pre-fill reduced onboarding by 40%.

Node.jsPostgreSQLReactNLPWCAG

How It Works

A WCAG-compliant healthcare platform for patient management and medical billing. Patients enter basic info; an NLP pipeline parses unstructured medical notes and insurance documents to auto-fill billing codes (ICD-10, CPT) and pre-populate forms. This reduced onboarding friction by 40% while maintaining HIPAA-compliant data handling throughout.

Key Skills Applied

  • NLP — Medical document parsing, entity extraction for billing codes
  • Node.js + PostgreSQL — HIPAA-compliant backend with encrypted data at rest
  • React — Accessible UI meeting WCAG 2.1 AA standards throughout
  • WCAG Compliance — Full accessibility audit with screen reader support and keyboard nav
04

Technical Arsenal

AI / LLM

GeminiOpenAI APILangChainRAG PipelinesAgentic AIHuggingFacePyTorchTensorFlow

Conversational AI

Dialogflow CX/ESIntent ModelingNLP PipelinesSTT / TTSIVRWebhook FulfillmentMulti-channel Bots

Languages

PythonTypeScriptJavaScriptC++SQLJava

Frameworks

ReactNode.jsFlaskDjangoExpressReduxReact NativeGraphQL

Data & Search

FireCrawlPerplexity APISupabasePostgreSQLMongoDBFirebaseMySQL

Cloud & DevOps

GCPAWSAzureDockerKubernetesGitHub ActionsJenkinsCI/CD
05

Let's build
something great.

Currently exploring opportunities to architect and build AI systems at scale.

Email saahithmatam@gmail.com LinkedIn saahithmatam GitHub saahithmatam
Click a planet to explore