Applied AI Engineer · End-to-end product builder

From first idea to shipped product.

I'm Priyam Sekra, an applied AI engineer and product lead. I handle the whole path: shaping the idea, scoping it, designing the architecture, building the voice and agentic systems, then deploying, launching and running them in production.

See the work Resume

Every phase, handled by me

01Ideation
02Scope and plan
03Architecture
04Build
05Test and evaluate
06Deploy and launch
07Run and improve
Live product

active users in Juno's first month: 500+active users in Juno's first month
teams at Bajaj Finserv's HackRx RAG hackathon: 4th of 20,000+teams at Bajaj Finserv's HackRx RAG hackathon
AI publications, one with a Best Research Paper Award: 4AI publications, one with a Best Research Paper Award

How I work

One engineer, the whole path.

I take a product through every phase myself, from the first idea to the release and what comes after. No hand-offs between the thinking and the building.

From ideaTo live product

01
Ideation
Start from the problem: who it is for, what it should do, and what would make it worth using.
- Problem framing
- Feature list
- User flows
02
Scope and plan
Turn the idea into a scope: what ships first, what waits, and how pricing and plans work.
- Scope documents
- Pricing and plans
- Milestones
03
Architecture
Design the system before writing it: agents, voice pipeline, data model and cloud services.
- System design
- Data models
- Vendor choices
04
Build
Write the AI services, the backend APIs, and the mobile and web apps people actually use.
- AI and voice agents
- Backend APIs
- Mobile and web apps
05
Test and evaluate
Run evaluation suites and acceptance tests against the real stack, so quality is measured.
- Eval suites
- Acceptance tests
- Latency benchmarks
06
Deploy and launch
Set up the cloud infrastructure, cut over to production, and release to both app stores.
- AWS infrastructure
- Store releases
- Production cutover
07
Run and improve
Add the admin tools, analytics and usage limits, then keep shipping based on real use.
- Admin consoles
- Analytics
- Regular releases
Result
Live product
On the App Store and Google Play, with real people using it.
Juno Tether

Live products

Shipped, and in people's hands.

Two products taken down that path and released on the App Store and Google Play.

01/Trugen AI · 2026

Live

Juno

iOS and Android

You ask. Juno calls.

Juno is an AI personal assistant for people in the US and Canada. Say what you need and it phones the business on a real line, talks to a real person, and comes back with a confirmed booking. It is also a companion you can chat with, speak to in real time, or see as a talking avatar.

What I worked on

Product and solution architecture for the whole assistant: chat, realtime voice and talking avatar
The outbound calling agent (clarify, research, call to book, confirm) on a realtime speech-to-speech model, with a shadow LLM that checks each live call against its goal and corrects course mid-call
Hermes, a multi-tenant agent service on AWS with Lambda dispatch, SQS routing, DynamoDB task state and watchdog recovery, so one deployment serves many users at once
Gmail and Google Calendar over OAuth, with timezone-aware scheduling on EventBridge
Vendor benchmarks on cost and time to first token, plus the architecture docs and handoff specs used by internal teams and external partners
Go-to-market: short-form content and paid-ad launch videos that took Juno to 500+ active users within a month

ElevenLabs
LiveKit
Deepgram
Telnyx
Mem0
AWS Lambda
SQS
DynamoDB
EventBridge
OAuth 2.0
Flutter
FastAPI

Download on theApp Store Get it onGoogle Play Website

Books restaurants, salons and spas, dentists, pet groomers, car servicing and activities
Tells the person who answers that it is an AI, every time
Waits for opening hours, calls back when asked, and phones to cancel if plans change
Proactive briefings and meeting prep from Gmail and Google Calendar

The avatar companion

The video-first side of Juno: choose an avatar and talk to it face to face, by voice or by chat. It remembers what matters between conversations and checks in with briefs and catch-ups.

Active users: 500+ in month one
Markets: US · Canada
Platforms: iOS · Android
Version 2.0: September 2026

02/Mentat Digital Solutions · 2025 – 2026

Live

Tether

iOS and Android

Where fathers and their children connect.

Tether is a relationship companion for fathers, sons, daughters and father figures. It pairs guided activities, conversation prompts and coaching with AI that helps both sides work through the moments that are hard to face.

What I worked on

The AI service behind journalling reflections, conflict mediation, Safe Space and activity recommendations, with LLM-based sentiment analysis and structured data capture
Safe Space end to end, from the API to the app screens
The personality assessment, and usage limits across the app, the API and the admin console
The analytics and content console the team runs the app from

Flutter
React Native
Python
FastAPI
PostgreSQL
LangGraph
OpenAI
Pipecat
React

Download on theApp Store Get it onGoogle Play Website Write-up

My Journey: a shared timeline of photos, video, voice notes and journal entries
Guided conversations and activities for the talks that keep getting put off
Safe Space: private, PIN-protected journalling with AI reflection
Two-sided conflict resolution, plus an AI coach you can talk to

An early design screen

The home screen from the first father and son build: a thought of the day, shortcuts to Safe Space and guided conversations, weekly engagement, and recent memories.

Platforms: iOS · Android
On the App Store: September 2026
Category: Lifestyle
First built for: Fathers and sons

More projects

Agents, voice and everything before.

All 15 projects

Agents & LLMs · 2026

Personal project

Vani

A WhatsApp second brain

Turns everyday WhatsApp messages, photos and voice notes into structured contacts, tasks, reminders and notes. Ask it “Who did I meet at that AI event?” and it searches long-term memory; each day it sends a planning summary. It also runs from a web portal and as an MCP server other agents can call.

TypeScript
Next.js
PostgreSQL
pgvector
MCP
Claude
WhatsApp Cloud API

Voice AI

Kailasa

Real-time speech-to-speech conversational AI

A low-latency voice assistant that chains speech recognition, a language model and text-to-speech into a fluid conversation, with interruption handling so it pauses and resumes the way a person would.

Voice bot
STT / TTS
LLM
Python
AWS

Visit site

Product

B.Tech capstone

Vizaro

GenAI and AR for personalised e-commerce

A framework that generates AR-ready product models with generative AI, so shoppers can see a product in their own space and get recommendations shaped to them.

Generative AI
WebAR
E-commerce
AWS
MongoDB

About

Applied AI engineer, product lead.

I'm an applied AI engineer and product lead based in Jaipur, India. I work directly with clients and business teams to scope ambiguous problems, build inside their stack, and take AI systems from prototype to production.

I'm comfortable owning a deployment end to end: integrating with existing systems over APIs, OAuth and event-driven AWS pipelines, monitoring agents in production, and explaining the trade-offs in latency, cost and reliability to engineers and business stakeholders alike.

Also startedAIProStack Solutions ResQNotify

Experience

Jan 2026 – Present
Trugen AIAI Product Lead, Juno AI
Remote
Own product and solution architecture for Juno: realtime voice agents, outbound AI calling, third-party integrations and go-to-market, growing it to 500+ active users in its first month.
May – Sep 2025
DHL Express IndiaSummer intern, process automation
Mumbai
Automated the Helpdesk India team's manual processes with Power Apps, Power Automate and Power BI. The workflows saved three to four hours a day.
Feb – Dec 2025
Mentat Digital SolutionsConsultant AI engineer and researcher
USA, remote
Turned emotional-mediation and healthcare workflows into AI tools: LLM-based sentiment analysis, structured data capture, journalling and workflow automation.
Mar – Dec 2025
NovusVistaConsultant conversational AI engineer
Jaipur
Took the client's voice AI assistant from requirements to production, with a fast RAG integration that made its knowledge base scale.
May 2024 – May 2025
QuibbleAIGenerative AI engineer intern
Jaipur
Led development and deployment of AWS-native conversational and WhatsApp AI agents in Python with Pipecat, owning analytics, automated deployments, production monitoring and project management.
Feb – May 2024
AIProStack SolutionsCo-founder, deep learning engineer
Jaipur
Built driver-monitoring models for tailgating, distraction, seatbelt and phone use, deployed on Raspberry Pi with live video streamed through Kafka for instant alerts.
Apr – Jul 2023
AAIS.AIMachine learning engineer
France, remote
Developed an AI-powered employee cybersecurity training platform, now deployed by AAIS.AI for enterprise clients.
2021 – 2026
MBA Tech and B.Tech, Computer Engineering
NMIMS Mukesh Patel School of Technology Management & Engineering, Mumbai
89%. Theses on smart technologies in logistics and transport, and on VIZARO, an AI and AR framework for e-commerce.

Toolbox

Voice AI and telephony

ElevenLabs
LiveKit
Pipecat
Daily WebRTC
Deepgram
Telnyx
Speech-to-speech models

LLMs and agents

Agno
LangGraph
LangChain
MCP
Mem0
RAG
Fine-tuning
RunPod

Integration and deployment

REST APIs
WebSockets
OAuth 2.0
AWS Lambda
SQS
EventBridge
DynamoDB
ECS
Docker
CI/CD

Programming

Python
SQL
Shell
TypeScript
FastAPI

Data science and ML

PyTorch
TensorFlow
Keras
scikit-learn
OpenCV
Pandas
Power BI
Streamlit

Product and delivery

Requirements scoping
Solution design
Rapid prototyping
Beta programs
Stakeholder communication
Go-to-market

Recognition

Tested outside the codebase.

Awards, research and the people I have built and published with.

Awards and achievements

2025
4th place, Bajaj Finserv HackRx
RAG hackathon, among 20,000+ teams
2025
NMIMS University Award
Best B.Tech final-year project (capstone)
2024
Best Research Paper Award, ICDTV 2030
For “SafeCam”, on AI-enabled driver analysis
2024
ICreate grant
Funding from the government-supported incubator for an AI driver assistance and behaviour detection system
2024
Semi-finalist, Eureka! 2024
IIT Bombay's business model competition
2024–25
Workshop host
A two-day chatbot bootcamp and two augmented reality workshops

Publications

01
Empathetic Intelligence: LLM-based Conversational AI Agent
International Journal of Intelligent Systems and Applications in Engineering (IJISAE)
Scopus-indexed
02
Enhancing Road Safety through AI-Enabled Driver Analysis
International Conference on Digital Technology Vision 2030 (ICDTV)
Best Research Paper Award
03
AI-Powered Employee Cyber Awareness System
Technische Sicherheit
Scopus-indexed
04
VIZARO: A Unified AI-AR Framework for Immersive and Personalized E-Commerce Experiences
International Conference on Sustainability, Innovation & Technology (ICSIT)
Under review

All four by Priyam Sekra and Vipul Bhatia.

Collaborators

Dr. Sandeep Pandey
University of Stuttgart, Germany
Driver behaviour detection and analysis
Sylvester Obafunwa
Industry advisor, Tata Consultancy Services, USA
Father and son relationship app
Jitin Doriya
CTO at QuibbleAI, IIT (BHU) Varanasi
Low-latency voice bot for large documents and restaurant menus
Dr. Sweta Soni
NMIMS MPSTME, Shirpur
Empathetic Intelligence: LLM-based conversational AI agent
Dr. Shailaja Rego
School of Business Management, NMIMS Mumbai
Review of smart technologies in logistics and transport
Dr. Mahendra Parihar
NMIMS MPSTME, Mumbai
Review of smart technologies in logistics and transport

Writing

All posts on Medium

Contact

Have something that needs to talk back?

I'm happy to talk about voice agents, LLM products, or a role where both matter.

priyam22rr@gmail.com LinkedIn GitHub

From first idea to shipped product.

One engineer, the whole path.

Ideation

Scope and plan

Architecture

Build

Test and evaluate

Deploy and launch

Run and improve

Live product