Who is Divyaprakash Dhurandhar?

Divyaprakash Dhurandhar is a Founding Full-Stack Engineer and Product Lead based in Gurugram, India. He holds an M.S. in Computer Engineering from Rutgers University (2020). He built TimekeeperX — a multi-tenant HRMS adopted by 20+ companies and 5,000+ employees — and agentic AI systems including voice agents using Gemini Live and Amazon Nova 2 Sonic. He is the creator of AutoShorts (Top 5 HackerNews, GitHub Copilot CLI Hackathon Runner-up), AgentLens (open-source AI agent visualization library), and TidyTUI (Rust TUI on crates.io and Arch AUR). Certified by Anthropic Academy in MCP, Agent Skills, and AI Fluency (2026).

What has Divyaprakash Dhurandhar built?

TimekeeperX: Multi-tenant HRMS on AWS ECS (Node.js, Next.js, MongoDB) — 20+ companies, 5,000+ employees. AutoShorts: GPU-accelerated video pipeline (Python, FFmpeg, OpenCV) — Top 5 HackerNews, 100+ GitHub stars, Copilot CLI Hackathon Runner-up. AgentLens: TypeScript library for visualizing AI agent browser actions (React + Vue adapters, npm published). TidyTUI: Rust TUI for Linux system maintenance — crates.io + Arch AUR. Badminton Vision Analytics: YOLOv8/v11 AI for shuttlecock and player tracking. Parameter Golf: 16MB LM for OpenAI Model Craft Challenge — novel architecture (BLT + MLA + KAN + BitNet), IEEE/arXiv paper.

What certifications does Divyaprakash have?

Anthropic Academy (2026): AI Fluency: Frameworks and Foundations, Introduction to Agent Skills, Introduction to Subagents, Introduction to MCP, Model Context Protocol: Advanced Topics. IBM/Coursera (2021): IBM Data Science Professional Certificate. deeplearning.ai/Coursera (2020): Deep Learning Specialization by Andrew Ng.

What roles is Divyaprakash Dhurandhar targeting?

Founding Engineer, AI/Agentic Systems Engineer, and Senior Full-Stack Engineer roles at Seed to Series B startups in the US and Europe (remote). Minimum $120K USD / €100K EUR. Equity is non-negotiable. Contact: divyaprakash.0426@gmail.com or linkedin.com/in/ddivyaprakash26.

Hi, I'm @divyaprakash,

Founding Engineer
who builds AI systems
with HITL + MCP + RAG

BuilderAI SystemsAutoShorts100+12

At AUM Group I built TimekeeperX from scratch: a multi-tenant HRMS now running for 20+ companies and 5,000+ employees on AWS. Then added voice AI agents that screen candidates autonomously over the phone.

This still feels like day one.

Founding roles. Harder AI problems. End-to-end ownership.

Ready for what's next.

My path What I build Let's talk Ask me

Work Experience

End-to-end ownership across architecture → engineering → deployment → adoption, with a strong bias for shipping fast and iterating in production.

Zero-to-One Shipping

Blank repo to production system in weeks. Founder mindset, full ownership.

Agentic AI Systems

Voice AI agents, LLM pipelines, autonomous screening, browser automation

Cloud Infrastructure

AWS ECS, Terraform, Docker, CI/CD — multi-tenant, production-grade

Full-Stack Engineering

TypeScript, Node.js, Next.js, React, MongoDB, REST & WebSocket APIs

Open Source Building

AgentLens, TidyTUI (crates.io + AUR), AutoShorts — real-world traction

Fast Prototyping

GPU pipelines, ML models, ComfyUI nodes — shipped quickly, iterated in prod

AUM Group

Gurugram, India

Founding Full-Stack Engineer & Product Lead

Nov 2023 – Present · SaaS / HR Tech

•Built TimekeeperX (timekeeperx.com) — multi-tenant HRMS from zero on AWS ECS (Node.js, Next.js, MongoDB)
•Adopted by 20+ companies, 5,000+ employees in production
•Built AI hiring agent: Amazon Nova 2 Sonic over SIP for autonomous candidate screening
•Built agentic HRMS navigation: Gemini Live API + AgentLens for voice-driven UI
•Led Quantamental Stock Prediction System: XGBoost/LightGBM ensemble + FinBERT sentiment
•Full infrastructure ownership: AWS ECS, Terraform, CI/CD, Docker
•Philosophy: ship fast, instrument everything, iterate in production

20+ companies across India

Manufacturing

IT Services

Logistics

Retail

Finance

Healthcare

timekeeperx.com

Source of Truth

TimekeeperX — Multi-Tenant HRMS

Production multi-tenant HRMS built from zero. Single codebase serving 20+ companies with full tenant isolation, role-based access, and real-time data. Live at timekeeperx.com.

Multi-tenant auth with full tenant isolation and RBAC
Attendance & leave management with approval workflows
Payroll engine with automated calculations and reports
AWS ECS deployment with Terraform IaC and auto-scaling
Gemini Live + AgentLens for voice-driven HRMS navigation
AI hiring agent via Amazon Nova 2 Sonic over SIP
Next.js dashboard with real-time updates and mobile support

Live at timekeeperx.com

20+

companies

5,000+

employees

~2yr

in production

Autonomous · Voice AI · SIP

AI Hiring Agent (Amazon Nova 2 Sonic)

Autonomous candidate screening over the phone. Amazon Nova 2 Sonic + SIP protocol — no human in the loop for first-round screening.

SIP integration: receives and makes real phone calls via VoIP
Amazon Nova 2 Sonic: real-time voice conversation with candidates
Structured screening: experience, skills, availability, culture fit
Automated report: full transcript + scoring → HRMS
HITL escalation: flags edge cases for human review

View code

Gemini Live · AgentLens · Voice UI

Agentic HRMS Navigation (Gemini Live)

Voice-driven navigation of TimekeeperX using Gemini Live API + AgentLens for real-time visualization of AI agent actions in the browser.

Gemini Live: real-time voice understanding + intent classification
AgentLens: cursor animations and element spotlight for agent actions
Natural language commands navigate HRMS workflows
WebSocket bridge between voice model and browser UI
Sub-200ms action latency in production

View code

Quantamental Stock Prediction

XGBoost/LightGBM + FinBERT — hybrid quant/sentiment model for stock signals

AUM Group internal

Multi-Tenant Infrastructure

AWS ECS + Terraform — full IaC for multi-tenant HRMS with auto-scaling

20+ tenants served

Automated CI/CD

Docker + GitHub Actions pipeline — zero-downtime deploys to AWS ECS

Zero downtime

Real-Time APIs

Node.js + WebSocket APIs for live attendance, payroll, and notifications

5,000+ users live

Open Source Contributions

ComfyUI node (MiDaS + Luma Photon) — novel-view synthesis in the creative AI ecosystem

DreamLayer AI

Remote

Backend Engineer (Open Source Contribution)

Jun – Aug 2025

Built ComfyUI custom node for novel-view synthesis: MiDaS depth estimation + Luma Photon Flash API. Dockerized dev environment with CI health checks.

Mine Gate India LLP

minegateindia.co.in Gurugram, India

Software Engineer

Nov 2022 – Nov 2023 · Drone Survey / GIS

Built photogrammetry pipeline using Colmap and OpenDroneMap — converting drone footage to 3D stockpile models for coal mine volume estimation. Set up on-premise Traccar GPS tracking system for 5-truck fleet in real time.

"Testimonial available on request."

Evergreen Technologies LLC

Quantitative Developer (Intern)

Oct 2019 – Jan 2020 · Fintech · NJ, USA

Contributed to IYKA Trading platform: MySQL DB-API integrations, RabbitMQ and Crossbar.io event-driven messaging architecture.

Rutgers Research (2017–2018)

R.A.D.I.C.A.L Lab: HPC interoperability for Radical-Pilot on XSEDE (Comet/Stampede). Cyber Physical Systems Lab: autonomous control for BlueROV2 underwater robot (ArduSub).

"Testimonial available on request."

Projects

divyaprakash0426

AI Systems Built

Production AI systems, not demos. Each solves a real problem at scale.

TimekeeperX AIproduction

Agentic HRMS navigation via Gemini Live + AgentLens

AI Hiring Agentproduction

Autonomous candidate screening via Amazon Nova 2 Sonic over SIP

AutoShortsopen source

GPU pipeline: gameplay → viral short-form content. Top 5 HackerNews.

AgentLensopen source

Browser library: visualize AI agent actions in real-time

Badminton Visionresearch

YOLOv8 shuttlecock + player tracking in real-time

Quant Systemprivate

XGBoost/LightGBM + FinBERT stock prediction pipeline

TimekeeperX

Flagship ProductLive · timekeeperx.com

Multi-tenant HRMS built end-to-end on AWS ECS (Node.js, Next.js, MongoDB). Adopted by 20+ companies and 5,000+ employees. Full attendance, payroll, leave management, and AI-powered navigation via Gemini Live + AgentLens.

Node.jsNext.jsMongoDBAWS ECSTerraformGemini Live

View prototype

AutoShorts

Open Source · Hackathon

GPU-accelerated pipeline that converts raw gameplay footage into viral short-form content. Top 5 HackerNews, 100+ GitHub stars, Runner-up GitHub Copilot CLI Global Hackathon 2026.

PythonFFmpegOpenCVNode.jsLLMGPU

View code100+

AgentLens

Open Source

Framework-agnostic browser library for visualizing AI agent actions in real-time. Cursor animations, element spotlight, and interaction tracking — works with React, Vue, and vanilla JS.

TypeScriptReactVueBrowser APIs

View code

TidyTUI

Open Source · crates.io · AUR

Rust terminal UI for Linux system cleanup. Published on crates.io and available on the Arch Linux AUR. Built with Ratatui.

RustRatatuiLinuxTUI

View code

Badminton Vision Analytics

Research · AI/CV

Real-time shuttlecock and player tracking using YOLOv8/v11. Shot classification, rally analytics, and performance metrics via computer vision.

YOLOv8PyTorchOpenCVPython

Parameter Golf

OpenAI Challenge

OpenAI Model Craft Challenge: 16MB language model trained in 10min on 8×H100s. Novel architecture: BLT tokenization + MLA attention + KAN + BitNet. IEEE/arXiv paper submitted.

PyTorchCUDABLTBitNetKAN

Anthropic-Certified AI Developer

Agentic Systems · MCP · Voice AI

Building production agentic systems is my daily work — from voice AI hiring agents over SIP to Gemini Live-powered HRMS navigation. Anthropic-certified across MCP, agents, and subagents.

›Voice AI: Built autonomous candidate screening agent using Amazon Nova 2 Sonic over SIP — full phone-round screening without human involvement
›Agentic HRMS: Integrated Gemini Live with AgentLens to enable voice-driven navigation of a production HRMS used by 5,000+ employees
›Open-source AI: AgentLens provides real-time visualization of AI agent actions (cursors, spotlights) for any framework
›Research AI: Parameter Golf — trained 16MB LM in 10min on 8×H100s with novel BLT + MLA + KAN + BitNet architecture

Introduction to Agent Skills Introduction to Subagents Introduction to MCP Model Context Protocol: Advanced Topics

Research & Recognition

Divyaprakash | Founding Engineer

@divayaprakashq

AutoShorts — GPU pipeline that converts raw gameplay into short-form videos. Watch the demo:

YouTube: youtube.com/watch?v=JZawIDjbxCg

YouTube Demo

@divyaprakash0426 · AutoShorts

"AutoShorts full demo — gameplay to viral short-form video in seconds"

100K+50K+500K+2M+Watch on YouTube

15K8K120K850KWatch on YouTube

Built AutoShorts: a GPU pipeline that converts raw gameplay footage into short-form content automatically. Top 5 HackerNews. Runner-up GitHub Copilot CLI Global Hackathon.... read more

34062r/SideProjectRead on Reddit

Shipped AgentLens: see what your AI agent sees in real time. Open-source browser library with cursor tracking, element spotlight, and React/Vue adapters.... read more

21038r/typescriptRead on Reddit

Built a voice AI hiring agent using Amazon Nova 2 Sonic over SIP. It screens candidates autonomously — full phone call, no human in the loop for round 1.... read more

18045r/MachineLearningRead on Reddit

AutoShorts hit Top 5 on HackerNews overnight — and won Runner-up at the GitHub Copilot CLI Global Hackathon. Here's what I built.... read more

200+40+Read on LinkedIn

Shipped AgentLens: an open-source library that lets you *see what your AI sees* — cursor animations, element spotlight, real-time agent visualization for any framework.... read more

150+25+Read on LinkedIn

Used an AI pipeline (career-ops) to evaluate 20+ job offers. Each one scored across 10 dimensions in seconds. The signal-to-noise ratio is remarkable.... read more

80+15+Read on LinkedIn

2026 · GitHub Copilot CLI Global Hackathon

Runner-up — AutoShorts

GPU-accelerated gameplay-to-shorts pipeline. Top 5 HackerNews, 100+ GitHub stars.

2026 · OpenAI Model Craft Challenge

Parameter Golf — 16MB LM in 10min

Novel architecture: BLT tokenization + MLA attention + KAN + BitNet. Trained on 8×H100s. IEEE/arXiv paper submitted.

2024 · Ministry of Railways, India

Train Drag Reduction Feasibility Study

OpenFOAM CFD analysis for Bird Eye View Pvt. Ltd. — feasibility study commissioned by Ministry of Railways.

2016 · E-Yantra National Robotics Competition, IIT Bombay

Top 11 / 5,000+ Teams

National robotics competition organized by IIT Bombay. Top 11 finish out of 5,000+ participating teams.

Anthropic-Certified AI Developer

MCP · Agents · Subagents · AI Fluency

Certified by Anthropic across the full agentic stack: AI Fluency Foundations, Agent Skills, Subagents, MCP (basic + advanced). Building production agentic systems — voice AI, autonomous agents, multi-step LLM pipelines — is my daily practice.

AI Fluency: Frameworks and Foundations Introduction to Agent Skills Introduction to Subagents Model Context Protocol: Advanced Topics

Education

2020 · Rutgers University

MS in Computer Engineering

GPA 3.5 · Coursework: HPC, Cyber-Physical Systems, ML. Research: Radical-Pilot HPC interoperability (R.A.D.I.C.A.L Lab), BlueROV2 underwater robot (CPS Lab).

2017 · R.C.O.E.M Nagpur