Technical Lead — AI Systems / Full-Stack

HEMENDRATRIPATHI

ARCHITECTUREBILLINGTEAMSREVENUE

Hemendra Tripathi is a technical lead for AI voice systems who takes products from first commit to paying customers — architecture, billing, teams, and the revenue they produce.

BaseUdaipur, IN
ServingUS / EU teams
CurrentCallin.io · Tech Lead
Proof1,500+ paying customers
REAL-TIME VOICE AI ✦MULTI-LLM ORCHESTRATION ✦USAGE-BASED BILLING ✦TWILIO / TELNYX / SIP ✦RAG · PINECONE · SUPABASE VECTOR ✦US / EU STAKEHOLDERS ✦ENTERPRISE HEALTHCARE · REAL ESTATE ✦REAL-TIME VOICE AI ✦MULTI-LLM ORCHESTRATION ✦USAGE-BASED BILLING ✦TWILIO / TELNYX / SIP ✦RAG · PINECONE · SUPABASE VECTOR ✦US / EU STAKEHOLDERS ✦ENTERPRISE HEALTHCARE · REAL ESTATE ✦
+
Paying customers on a platform he architected
−%
LLM inference cost via model routing
ms
Typical voice TTFT after orchestration work
+
Prospects personally closed into paying accounts
(01)

Case Study

Featured case study

How we made voice agents cheap enough to scale — and fast enough that humans stayed on the line.

As Technical Lead I owned the architecture, vendor spend, and billing system for an AI voice platform that grew past 1,500 paying customers — including medical and real-estate enterprise accounts — while cutting LLM cost 20% and keeping billing disputes near zero.

callin.io ↗
The problem
  • Every turn of a voice agent is a race: if the model thinks too long, the caller hangs up.
  • Using one large model for every utterance burned margin on greetings and confirmations.
  • Minute-based billing with rollovers was creating support tickets — and eroding trust with the accounts that mattered most.
1,500+
Paying customers
−20%
LLM inference cost
~420ms
Voice TTFT (typical)
~0
Billing disputes since launch
What I built

Complexity-aware model routing

Classify each turn — greeting, FAQ, scheduling, objection — and route to the smallest model that can finish the job. Large models only for hard reasoning.

Semantic cache + concurrent prompts

Cache high-frequency intents; fire retrieval and response scaffolds in parallel so time-to-first-token drops before the caller notices silence.

Dual-carrier telephony

Twilio + Telnyx with SIP fallback. Fail over without dropping the call; keep audio streaming over WebSockets under load.

Billing as a product surface

Minute tracking with rollover ledgers precise enough that disputes became rare. Stripe subscriptions wired to actual usage, not estimates.

ReactNode.jsSupabaseTwilioTelnyxStripeElevenLabsCartesiaRedis

“Hemendra is the rare engineer who can rewrite the voice pipeline before lunch and close an enterprise prospect after dinner. He treats infrastructure cost like product debt — and it shows in the margins.”

(02)

Selected Work

01

Callin.io

IN PRODUCTION
Technical Lead — architecture, billing, voice pipeline

AI voice-calling SaaS scaled to 1,500+ paying customers and enterprise accounts. Low-latency telephony, multi-LLM orchestration, and minute-based billing with negligible disputes since launch.

React / Node.js / Supabase / Twilio / Telnyx / StripeFull case study →
02

CondoMail

LIVE
Product architecture — multi-provider email sync

AI agents that sort, draft, and auto-reply across providers. Live with early adopters — including high-volume Amazon sellers running inbox workflows on it.

React / Node.js / Supabase / Stripe / Firebase
03

Realead

BETA
Full-stack · mobile · AI calling flows

Connects lead sources, builds a business profile, and places AI qualification + follow-up calls for property leads. Final beta ahead of release — the conversational model behind the demo below.

React Native / NestJS / Supabase / Stripe
04

Sunria & FinTech Accounts

SHIPPED
Freelance — end-to-end delivery

Pan-India farm management with field-to-warehouse sync, plus a financial dashboard with automated reconciliation — 30% fewer accounting errors, 15+ staff-hours saved weekly.

Laravel / Flutter / MERN / CI-CD
(03)

The Demo Is the Résumé

I build AI agents that make real phone calls for a living. This one runs on the same conversational patterns as production voice agents — except its lead-qualification target is you.

Answer the call. Ask about numbers, stack, or why hire him. The lead file builds the way Realead qualifies property leads in the field.

  • → Prefer skimming? Read the case study first.
  • → Prefer proof? Finish the call. Get the summary.
Live demo — Recruiting-qualification line
HT
● Incoming call

Hemendra’s AI Agent

Built on the same stack as his production voice agents. Its only job: qualify you as a hiring lead.

(04)

How I Think

01

Latency is the product

In voice AI, silence is a bug. Every architectural choice — caching, routing, carrier failover — exists to keep the human from hanging up.

02

Pay for intelligence only when you need it

A confirmation doesn't deserve a frontier model. Route by complexity. Your CFO will notice. So will your p95.

03

Billing that doesn't create tickets

If customers argue about invoices, the product is unfinished. Usage ledgers should be boringly correct.

04

Own the stack's P&L

Architecture without vendor spend ownership is theater. I hire, I ship, and I know what the infra bill was last Tuesday.

(05)

Signal

“We evaluated three voice vendors. Callin's agents were the only ones our clinic staff didn't hang up on — and the only ones whose invoices we didn't audit line by line.”
Priya NairDirector of Operations, Meridian Family Clinics
“He hired half my eng team, set the roadmap, and still jumped on customer calls. That's not a contractor. That's an owner.”
Rohit MalhotraFounder, Appspundit Infotech
“Curriculum he redesigned moved our placement rate up over 40%. Students left knowing how to ship, not just pass exams.”
Ananya SharmaProgram Head, Aimers Institute
(06)

Experience

OCT 2024 — PRESENT

Technical Lead / Full-Stack DeveloperAppspundit Infotech · Callin.io

  • De facto Technical Lead for a 5-engineer team — architecture, vendor & infrastructure spend, hiring, product roadmaps, reporting directly to the founder.
  • Scaled the platform to 1,500+ paying customers; expanded into CondoMail and Realead on a shared multi-LLM architecture.
  • Cut LLM inference costs 20%; personally converted 30+ prospects into long-term paying accounts across healthcare and real estate.
2022 — 2024

Freelance Full-Stack DeveloperRemote · fintech, retail, logistics

  • Delivered 10+ end-to-end applications across MERN, Django, Flask, and Laravel.
  • Built a fintech accounts system for an MCA-registered firm — automated reconciliation saving 15+ staff-hours per week.
  • Improved API performance 25% and implemented zero-downtime CI/CD pipelines.
2021 — 2023

Technical InstructorAimers Institute & VT College

  • Mentored 150+ students in Python, Django, MERN, and Flutter through project-based learning.
  • Redesigned the curriculum to industry needs — student placements rose 42%.
  • Supervised 30+ capstone projects: version control, API design, UI craft.
MCA — Rajasthan Vidyapeeth (exp. 2026)BCA — Mohanlal Sukhadia University (2022)English — fluent · Hindi — native
(07)

Capabilities

AI & Voice Systems

  • Multi-LLM orchestration — routing by complexity & cost
  • RAG pipelines — Pinecone, Supabase Vector
  • Voice cloning — ElevenLabs, Cartesia
  • Real-time telephony — Twilio, Telnyx, SIP

Product & Revenue

  • Usage-based billing architecture
  • Stripe subscriptions & invoicing
  • Pricing design & infra cost optimization
  • Client acquisition & retention

Full-Stack Engineering

  • React, Next.js, React Native
  • Node.js, Express, NestJS, Laravel, Python
  • PostgreSQL (Supabase), Redis
  • Event-driven systems, REST, microservices

Cloud & Leadership

  • Docker, AWS, CI/CD pipelines
  • System design & architecture decisions
  • Team leadership — hiring & mentorship
  • US/EU stakeholder coordination
(08)

Off the Record

Based in Udaipur, shipping for the US and Europe. I got here by teaching 150+ students to code, freelancing across four frameworks, and rebuilding a voice-AI platform until 1,500 companies paid for it. I like systems that are boring, fast, and profitable — and teams that own what they build.

GitHub ↗LinkedIn ↗Résumé ↓
(09) — Contact

HIRE ME.

Looking for a technical lead who has already shipped AI products into revenue — not someone who will learn voice AI on your dime. Open to technical-lead / senior full-stack roles and select freelance. Replies within 24 hours.

GitHub ↗LinkedIn ↗Résumé ↓Case study ↑