Available for opportunities

Narayanan Satyamurthy

18+ year IT enterprise leader turned AI engineer — building production voice agents, multi-tenant SaaS platforms, and LLM-powered systems from the ground up.

18+Years Enterprise IT
8AI Products Built
1Hardware Prototype Co-Invented
$750MPlatform Managed

Enterprise Leader.
AI Builder.

I spent 18+ years as an IT strategy and service delivery executive — managing Bayer's $750M FieldView platform across 23 countries, leading 100+ person global teams at Cognizant, and driving SLA compliance from 40% to 85% at scale. Along the way I led Cloud FinOps initiatives that cut enterprise cloud spend materially — right-sizing compute, driving Reserved Instance coverage, and building cost-attribution frameworks across 40+ applications.

In 2025 I turned that operational depth toward building AI products. I'm now the sole founder and engineer behind four production AI SaaS platforms, each leveraging different AI capabilities — LLM reasoning, voice agents, vector embeddings, and real-time audio processing.

I don't just integrate APIs. I architect multi-tenant systems with proper RBAC, encrypted PII, async workers, CI test suites, and real paying customers. Based in Knoxville, TN.

B.Sc. in Electronics (dual major, Electronics & Management Information Systems).

🎙️

Voice AI

Vapi.ai, Retell AI, Twilio, Deepgram, ElevenLabs

🧠

LLM Engineering

GPT-4o, embeddings, RAG, structured outputs, prompt design

🏗️

SaaS Architecture

Multi-tenant, Supabase RLS, JWT auth, Railway, Docker

🏢

Enterprise Delivery

ITIL, SAFe Agile, SRE, ITSM, P&L, C-level leadership

AI Products Built

Eight AI products — from multi-tenant SaaS voice platforms to personal tools — each tackling a different problem with voice agents, LLM reasoning, and applied engineering.

📞

KnoxCalls NightWatch

🔗 knoxcalls.com

Multi-tenant AI after-hours answering service for home service businesses (HVAC, plumbing, electrical). An AI voice agent triages calls — dispatching technicians for emergencies via SMS/voice, logging routine leads to a CRM dashboard.

  • GPT-4o triage engine classifies Emergency vs. Routine in real time; Vapi.ai voice pipeline with Deepgram STT + ElevenLabs TTS
  • Bulldog Dispatch — async Twilio escalation loop pages on-call technicians via SMS and voice until acknowledged
  • Multi-tenant Supabase CRM with full call transcripts, audio recordings, lead pipeline, and modular add-on feature gating
  • Full observability stack — Grafana Cloud (Prometheus/Loki/Alloy), GlitchTip, UptimeRobot — with tiered incident-response matrices mapping SLO breaches to webhook alerts
  • JIRA + Confluence integrated for project tracking; 98%+ SLA compliance maintained in production
GPT-4o Vapi.ai Twilio Deepgram ElevenLabs FastAPI React 18 Supabase Railway
🍽️

Voco — AI Restaurant Ordering

🔗 voco.goldenkall.com

Multi-tenant SaaS AI voice ordering platform for restaurants. Handles up to 20 concurrent inbound calls — recognizes repeat callers, reads the menu, captures orders, and routes directly to the kitchen with zero human intervention.

  • Caller recognition CRM — identifies repeat customers by phone, surfaces order history; personalized interactions drive loyalty
  • GPT-4o Vision OCR menu import — restaurants photograph menus and AI extracts all items, prices & categories in <60s
  • Square Orders API + Toast POS integration; multi-channel kitchen routing: Supabase Realtime, email, Twilio SMS/WhatsApp, ESC/POS cloud printers
  • Stripe hybrid billing: monthly subscription base + per-minute overage metering; Turborepo monorepo (Fastify API + Next.js 15 web)
Retell AI GPT-4o Vision Square POS Toast POS Node.js/Fastify Next.js 15 Stripe Supabase
💼

JobQual — AI Recruiting Voice Agent

🔗 jobqual.goldenkall.com

Multi-tenant AI voice agent for recruiting agencies. Conducts inbound and outbound candidate screening calls, scores against per-role rubrics, books interviews, syncs to ATS — with vector embedding matching and a vendor resume portal.

  • AI matching engine: OpenAI text-embedding-3-small (1536-dim) + cosine similarity + GPT-4o LLM reranker; pre-computed fast path delivers sub-10ms cached results
  • Multi-skill voice agent framework: pluggable skills for screening, scheduling, reference checks, re-engagement; per-call system prompt assembler
  • Vendor portal: public token-based resume upload (PDF/DOCX/TXT); LLM contact extraction; background ingest with pgvector embedding
  • CRM: Fernet PII encryption, DNC enforcement, campaign management, ATS export (Greenhouse + HMAC-SHA256 webhook). 353/353 backend tests green.
Vector Embeddings Vapi + Twilio LLM Reranker ATS Integration FastAPI React/Vite pgvector Fernet PII
🛡️

CallSentry — AI Scam Detection

🔗 github.com/NanaHarith/callsentry

AI-powered scam detection app protecting elderly users from phone and SMS fraud. Real-time call analysis with GPT-4o-mini + Whisper; native Android app intercepts live calls; family members get instant alerts when threats are detected.

  • Real-time scam detection via WebSocket: Whisper STT converts audio → GPT-4o-mini classifies threat level (urgency language, gift card requests, authority impersonation)
  • Native Android (Kotlin) app using Telecom API to intercept live calls; push alerts dispatched to family members on threat detection
  • Family oversight dashboard: links multiple accounts, shows protection status, call/SMS history with full transcripts and threat scores
GPT-4o-mini Whisper STT Android Kotlin WebSocket FastAPI Firebase React 19
🗣️

Ringer — AI Voice Receptionist

🔗 github.com/NanaHarith/ringer

Persona-driven AI phone receptionist that answers calls as the owner's stand-in — paired with a companion Builder Agent that runs coaching calls with the owner to enrich the AI's persona and knowledge over time.

  • Builder Agent onboarding — structured coaching calls progressively build out the receptionist's persona, tone, and business knowledge
  • Layered prompt composition assembles a per-call system prompt from persona, business context, and live conversation state
  • Voiceprint recognition for returning callers and semantic/vector search over the owner's knowledge base for accurate answers
  • Gamification layer scores and rewards owners for completing persona-building milestones
Voice AI Builder Agent Vector Search Voiceprint ID FastAPI
📤

crosst — AI Outbound Voice Dialer

🔗 github.com/NanaHarith/crosst

Multi-tenant AI voice dialer platform for outbound calling campaigns. Tenants activate Skills and premium Add-ons à la carte, with calls routed across multiple telephony providers rather than locked to one carrier stack.

  • Multi-provider telephony routing abstracts across several voice/carrier backends for redundancy and cost control
  • Skills + premium Add-ons marketplace — tenants opt into campaign capabilities as needed
  • Built for outbound calling at scale: campaign management and compliance-aware dialing
  • Multi-tenant SaaS architecture with per-tenant data isolation
Multi-Provider Telephony Outbound Campaigns Multi-Tenant Skills Marketplace
🎯

HireMe — AI Job-Search Tool

🔗 github.com/NanaHarith/HireMe

Personal job-search tool that aggregates listings from ATS boards and other sources, scores each one's fit against a CV, and drafts tailored application materials — built to cut down the manual grind of applying.

  • Pulls job listings from ATS boards and other sources into a single feed
  • LLM-based fit scoring against the candidate's CV
  • Auto-drafts tailored CVs and cover letters per application
LLM Matching ATS Aggregation CV Generation
🧩

manas — Cognitive Exercise App

🔗 github.com/NanaHarith/manas

A cognitive exercise application — structured exercises designed to build and track mental sharpness over time. Actively in development.

  • Structured cognitive-training exercises with progress tracking
  • Actively evolving — newest addition to the product lineup
Cognitive Training In Development

Novel Field Instrument — Prototyped & Validated

Hardware meets biology — a novel field instrument co-invented with a postdoctoral researcher. A formal patentability search has been completed by patent counsel; the prototype itself is built, tested, and field-validated.

🕷️

Field-Deployable, Self-Calibrating Vibration Analysis System for Spider Web Characterization

📄 Patentability Search Completed  ·  Co-invented with Dr. Bharat Parthasarathy (SSSIHL, India)

Conventional spider web vibration research relies on Laser Doppler Vibrometers costing tens of thousands of dollars and requiring controlled lab conditions. This invention delivers equivalent measurements in the field — at 1/100th the cost — using off-the-shelf components combined in a novel configuration. Functional prototype built, tested, and field-validated.

  • Novel hardware integration: ESP32 microcontroller + MPU6050 accelerometer + mini vibrator motor — independent vibrator and sensor modules allow spatial frequency mapping at varying distances across the web, impossible with any existing system
  • Self-calibration system: Real-time environmental baseline measurement, automatic orientation compensation, and dynamic noise filtering — addresses the critical limitation of field sensors lacking real-time compensation
  • Mobile-first standalone operation: ESP32 creates its own WiFi access point; web-based interface lets a field biologist run full calibration and data capture with only a smartphone — no laptop, no internet
  • Configurable: Sampling rates 250–3000 Hz; validated at 500/1000 Hz for spider web resonance; automated SD card data management with real-time validation
  • Status: A formal patentability search was conducted by patent counsel. The project is not currently being pursued for patent filing and remains an internally validated prototype.
ESP32 Firmware MPU6050 Accelerometer IoT / Embedded Systems Real-Time Calibration Mobile WiFi Interface Field Biology Instrumentation

18+ Years of Enterprise Delivery

Global IT leadership across Fortune 500 clients, managing platforms, people, and P&L at scale.

🏥
CVS Health — Remote, Knoxville TN Nov 2024 – Present
Sr. Service Delivery Manager (Contract) – Applications Support
  • End-to-end infrastructure & application service delivery for 42+ enterprise Tier 1/2 platforms
  • Sustained 98%+ SLA compliance; operational oversight for call center IVR infrastructure
  • Migrated 14 applications from MSPs to in-house teams (~4 weeks/app), cutting delivery costs ~20%
  • Built internal LLM tooling that auto-updates runbooks from ServiceNow tickets and RCAs; deployed Teams-integrated self-serve chatbots
  • Executive-level reporting on infrastructure health, risk, and delivery performance
🌱
Bayer Crop Science — Remote, Knoxville TN Aug 2022 – Jun 2024
Senior Engineering Manager – Application & Infrastructure Operations
  • Led global 24×7 operations for FieldView, Bayer's $750M+ digital farming platform — 23 countries, 400+ microservices, 99.9% availability
  • Reduced MTTR from 96 hours to 90 minutes; zero repeat major incidents; improved SLA compliance from 40% to 85%
  • Reduced operational disruptions 60% during critical planting/harvest seasons via infrastructure reliability strategy
  • Cloud FinOps: right-sized compute and re-tuned auto-scaling to cut EC2/ELB spend ~30%; drove RI/Savings Plans coverage to 70%+, reducing baseline compute costs 35%
  • Core member of the working group leading Bayer's org-wide SAFe (Scaled Agile Framework) transformation — guided Application Support and 400+ engineers across multiple Agile Release Trains through the change; recognized with Bayer's "Better Because of You" award for the organizational change management effort
💻
Cognizant Technology Solutions — South Plainfield, NJ Oct 2015 – Aug 2022
Associate Service Delivery Director – App Dev & Maintenance
  • Led 65+ person global teams (US, Poland, India) delivering for Fortune 500 ecommerce & retail clients
  • Grew account revenue from $9M to $13M TCV; supported $2B+ holiday retail revenue with zero Sev-1 incidents
  • Led rapid 6-week infrastructure transition for major consumer electronics retailer; reduced outages 60%
  • Built enterprise-wide cloud cost-tagging taxonomy (100% spend attribution across 40+ apps) and a FinOps show-back framework that surfaced $500K+ in annualized savings; storage tiering/lifecycle cleanup cut storage spend ~25%
🏦
JP Morgan Chase — Bangalore, India Jan 2006 – Jun 2012
Assistant Vice President – Global Technology & Infrastructure
  • Led IAM for Investment Bank and Commercial Bank lines; managed 42+ professionals
  • Led India technology integration for Bear Stearns, Washington Mutual, and Bank of New York M&A
  • Drove 300%+ increase in India operations footprint; implemented automation yielding 36 FTE efficiency gains

Skills & Stack

The full technology footprint across AI, voice, backend, frontend, and enterprise tooling.

AI & LLM

  • OpenAI GPT-4o / GPT-4o-mini
  • GPT-4o Vision (OCR)
  • Whisper STT
  • text-embedding-3-small (RAG)
  • LLM prompt engineering
  • Structured outputs (JSON schema)

Voice & Telephony

  • Vapi.ai — voice agent orchestration
  • Retell AI — conversational voice
  • Twilio — SMS, voice, SIP
  • Deepgram — real-time STT
  • ElevenLabs / Cartesia TTS
  • SignalWire / WebSocket audio

Backend

  • Python 3.11/3.12 + FastAPI
  • Node.js + Fastify v5
  • Supabase (PostgreSQL + pgvector + RLS)
  • Firebase Firestore + Storage
  • Background workers, async queues
  • Fernet encryption, JWT auth

Frontend

  • React 18/19 + Vite + Tailwind
  • Next.js 15 (App Router)
  • Turborepo monorepo
  • Shadcn/UI, Supabase Realtime
  • Android (Kotlin) — Telecom API
  • React Native / Expo

Cloud & DevOps

  • Railway (Dockerized services)
  • AWS (Certified Cloud Practitioner)
  • Cloud FinOps — cost tagging, right-sizing, RI/Savings Plans
  • Stripe — subscription + metering
  • Resend, SendGrid (transactional email)
  • Square Orders API, Toast POS
  • Docker + CI/CD pipelines

Enterprise & Observability

  • ITIL v4 · SAFe Agile · SRE
  • Jira / Confluence
  • AppDynamics, Datadog
  • Splunk, ELK Stack
  • ServiceNow ITSM
  • Greenhouse ATS integration

Let's Work Together

Open to senior AI engineering roles, CTO/VP Engineering positions, consulting engagements, and partnership discussions around my SaaS products.

Ask about Nana
AI-generated answers, grounded in this profile
Hi — I can answer questions about Nana's background, projects, and experience, based on what's on this page. What would you like to know?
Answers are AI-generated from Nana's profile and may be incomplete — for anything else, use the contact links above.