Abhishek

Full Stack Engineer

NestJSNext.jsPostgreSQLMongoDBRedisQueues

2+ years building real-time systems, AI-integrated backends, and async workflows. Most of what I've built has gone into production and stayed there.

Platforms in production ·Healthcare ·EdTech ·Hiring-Tech ·Mithila Stack

Experience

Where I've worked

July 2026 - Present

Mithila Stack

Project Manager

In July 2026 I was promoted to Project Manager for the company, and I took it on alongside my engineering work, not instead of it. I now look after several projects at the same time and a team of 15-20 people across them.

A normal week is mostly planning what each team works on next, splitting work so people aren't waiting on each other, and talking to the team often enough that problems come up early instead of the day before a release.

I still write and review code. Staying close to it helps me give timelines I actually believe, and notice when something is harder than it looked in planning.

March 2024 - Present

Mithila Stack

Full Stack Engineer

Working on the backend of multiple production systems across different domains — hiring-tech, healthtech, edtech, and food-tech.

Built and shipped four products from scratch — Meleys (AI interview and assessment platform), MARS (medication adherence platform with AI-driven patient follow-ups), iPariksha (online examination system with automated evaluation), and Foodam (a city-level tiffin and food ordering service). Owned the backend end to end on all of them — API design, database, async workflows, third-party integrations, and deployment.

Day to day I work with Node.js, NestJS, Express.js, Fastify, PostgreSQL, MongoDB, Redis, and BullMQ. Most of the work has involved real-time systems, background job processing, and integrating AI services into practical workflows.

Handle deployment and infrastructure using Docker and Jenkins CI/CD pipelines on AWS. Work with HIPAA-compliant infrastructure for healthcare applications, managing sensitive PHI data with proper security controls and isolated services.

Backend

Node.jsNestJSExpress.jsFastifyPostgreSQLMongoDBRedisBullMQ

DevOps & Cloud

DockerJenkinsAWSCI/CDLinux

Frontend

Next.jsReactTypeScriptTailwind CSS

Selected work

Platforms I've built. All in production.

Meleys

Hiring-Techmeleys.ai

An AI-powered interview and evaluation platform with real-time proctoring. Handles candidate screening, live interviews, and automated scoring against role-specific criteria.

interviews across 1,000+ candidates: 2,500+
concurrent live sessions: 100-150
event capture latency: ~40-50ms
AI scoring accuracy: ~85%

Built a real-time interview system with event-driven architecture for low-latency activity capture. The scoring engine evaluates candidates against custom criteria per role, and the system handles multiple concurrent sessions without performance drops.

NestJSPostgreSQLRedisBullMQWebRTCGemini AI

AI Clinic

Healthcareai-clinic.tech

India's default medication adherence platform with multiple AI agents working 24/7. Built multiple real-time conversational agents on LiveKit that connect hospitals with patients after surgery or discharge - monitoring recovery, delivering care instructions, and managing medication schedules. Integrated with MARS system to support hospital internal operations.

Architected a multi-agent system on LiveKit for real-time patient engagement across multiple hospitals. Each agent handles specific care moments: pre-surgery prep, medication adherence, post-surgery monitoring, prescription fulfillment, and appointment management. Hospitals can choose from multiple intervention agents based on patient needs. Integrated WhatsApp API to deliver AI doctor avatar videos for medication reminders. Built event-driven workflows with Python and Node.js for intelligent nudges. Seamlessly integrated with MARS (hospital management system) to unify patient engagement with internal workflows.

Node.jsPythonLiveKitMongoDBAzure AID-IDWhatsApp APIDockerJenkinsAWSBullMQ

MARS

Healthcaremars.ai-clinic.tech

Medication Adherence & Reminder System - A comprehensive hospital management platform with AI-powered patient engagement. Handles OPD operations, daily appointments, prescription management, and automated patient reminders through customizable AI agents.

patients on WhatsApp reminders: 2,000+
reminders sent every month: 25,000+
delivery success: 90%
better on-time medication adherence: ~50%

Built a complete hospital management system with OPD module supporting daily appointment scheduling, prescription creation and management, and patient records. Implemented background processing for time-based, retry-safe task execution at scale. Developed an AI agent system where hospitals can select from multiple intervention agents for post-discharge patient follow-up, medication awareness, and care instruction delivery. The customized doctor avatar calls patients, delivers medication reminders with precautions, and handles follow-ups - fully automated end to end. Enforced role-based access control and secure handling of sensitive healthcare data with isolated services and scalable storage.

Node.jsMongoDBAzure AID-IDFFmpegDockerBullMQRedis

iPariksha

EdTechipariksha.com

An examination platform for conducting large-scale online tests with automated evaluation. Handles test creation, submission processing, result generation, and integrated payment flows.

exams conducted: 1,500+
candidates: 5,000+
from submission to results: 5-15 min

Designed backend systems for large-scale examinations. Structured workflows for test creation, submission handling, and result generation. Integrated Razorpay payment and payout flow into the core system while maintaining reliability and enforcing clear role boundaries.

NestJSPostgreSQLRedisBullMQRazorpay

TiffinDost

FoodTech

A meal subscription aggregator connecting users to local kitchens for daily tiffin delivery. Handles subscription management, kitchen partnerships, and delivery coordination.

Subscription logic is deceptively tricky. Pause, resume, skip, swap kitchen - all without breaking billing state. Built with a small team and shipped to production in weeks.

TypeScriptMySQLRedisAWS EC2WebSockets

How I work

Backend engineering is about decisions, not just code.

Things break, plan for it

Queues have retries. Tasks have fallbacks. Auth has layers. I build assuming something will go wrong, because in production, it usually does.

Get the data model right first

Schema decisions shape everything else. I spend time on data models before writing routes - fixing a bad schema later costs way more.

Ship it, then make it better

A working system teaches you more than a design doc. I get something real out fast, then improve based on what actually happens.

Languages: TypeScript, JavaScript, Python

Backend: Node.js, NestJS, Fastify, Express.js

Databases: MongoDB, PostgreSQL, MySQL

Frontend: Next.js

DevOps & Cloud: Docker, Jenkins, AWS EC2, AWS S3, CI/CD, Linux

Systems: Redis, BullMQ, WebSockets, WebRTC, LiveKit, Media Processing

AI & ML: Gemini AI, Azure AI, D-ID, OpenAI, Machine Learning

Integrations: WhatsApp API, Razorpay, FFmpeg

Have a hard problem?abhias.dev@gmail.com

Connectlinkedin.com/in/AbhiMehrr